ICAD Inc (ICAD) 2018 Q4 法說會逐字稿

Greetings.

Welcome to the iCAD, Inc.

Fourth Quarter and Full Year 2018 Earnings Conference Call.

(Operator Instructions) Please note this conference is being recorded.

I will now turn the conference over to your host, Jeremy Feffer, Managing Director at LifeSci Advisors.

Mr. Feffer, you may begin.

Thank you, Omer, and good afternoon, everyone.

Thank you for participating in today's call.

Joining me from iCAD today are Michael Klein, Chairman and Chief Executive Officer; Stacey Stevens, Chief Strategy and Commercial Officer; and Scott Areglado, Interim Chief Financial Officer.

Earlier this afternoon, iCAD announced financial results for the 3 months and full year ending December 31, 2018.

Before we begin, I would like to caution that comments made during the conference call by management that contain forward-looking statements involve risks and uncertainties regarding the operations and future results of iCAD.

I encourage you to review the company's filings with the SEC, including, without limitation, the company's Forms 10-Q and 10-K, which identify specific risk factors that may cause actual results or events to differ materially from those described in the forward-looking statement.

Furthermore, the content of this conference call contains time-sensitive information that is accurate only as of the date of this live broadcast, March 18, 2018 -- 2019, excuse me.

iCAD undertakes no obligation to revise or update any statements to reflect events or circumstances after the date of this conference call.

With that said, it's my pleasure to turn the call over to Michael Klein.

Mike?

Thank you, Jeremy.

Good afternoon, everyone, and thank you for joining us today.

I am pleased to have the opportunity to speak with you today and update you on our recent strategic progress and lay out some of our future plans.

With me today are Scott Areglado, our Interim CFO and Corporate Controller; and Stacey Stevens, our Chief Strategy and Commercial Officer.

On our last call, my first as CEO, I laid out my vision for iCAD.

In detection, I firmly believe that the company's unique AI technology is transforming cancer detection and reinforcing iCAD's leadership in computer-aided detection or CAD, which we have now evolved into a robust AI offering.

With the recent FDA approval in December and the subsequent U.S. commercial launch of ProFound AI, our latest deep-learning cancer detection software solution for digital breast tomosynthesis or DBT, we are facilitating vast improvements in workflow, reading time as well as clinical efficacy.

Achieving a quantum leap in sensitivity while improving specificity or fewer false positives represents a clear breakthrough in AI capability.

The feedback from our initial customers now actively using ProFound has been very positive.

AI has been -- is now on its way to achieving -- with iCAD, is on its way to achieving our goals with this exciting offering.

As I said on our last call, we intend to lead the way in 3D AI and play a leadership and perhaps even an ambassadorial role, as we navigate both near-term opportunities and move towards new horizons of highly personalized detection tools and risk assessment.

As an extension of this, we are positioning and growing iCAD in a different way.

Up till very recently, we have seen and have been seen as a breast cancer or women's health company that deploys AI software technology.

Today, we see ourselves moving towards a broader-based clinical AI company that has gone to market in the breast cancer area first, but can deploy the same AI capabilities either on our own, or more likely, with channel partners, in addition to medical imaging and disease diagnostics' new frontier.

With the only FDA-approved product for 3D tomosynthesis, we started the initial build-out of our commercial infrastructure in Q4.

As of today, we have now doubled our field infrastructure from less than 10 sales and service personnel at the beginning of Q4 to now over 21 carefully selected clinical AI consultants, inclusive of robust service, support, install personnel.

This allows us to educate customers and to continually improve our offering and proliferate ProFound AI technology.

We may add 2 to 3 additional field resources to this infrastructure around mid-year as installations ramp, and in doing so, we'll nearly triple our headcount when compared to Q3 of 2018.

For our Q1 and Q2 adoption plans and revenue target, we have the full talent set we need, targeted and ready and onboard today.

We are executing our plans for a controlled yet scalable launch.

We are convinced that a direct sales model while working collaboratively with partners is the only way to optimize the true value of our novel technology, train many nascent adopters and customers' evolving clinical needs.

We hope to meet them even better with this infrastructure.

Even as this commercial transition was unfolding in late Q3 and then accelerated in Q1, our Q4 sequential sales -- total sales as compared to Q3 grew by 12%.

AI revenue led the way with an increase of $1 million or 26% versus Q3 sales in 2019.

We are very pleased with the continued growth of our intraoperative radiation therapy, IORT offering, within with Xoft, had a total year-over-year basis of growth of 22% over 2017 on therapy products revenue.

The benefit of a 10-minute real-time procedure during lumpectomy is now moving on a positive trajectory.

Simultaneously in Q4, we completed a $7 million financing at the end of December, which Scott will discuss further shortly.

This was done to support the aforementioned commercial infrastructure build-out but also to accelerate the development of new clinical AI technology.

I will add some color on this in my opening comments.

We have now achieved our initial Phase I sales and installation goal of having customer-acquired systems in use in targeted major U.S. regions.

Regional adoption is an essential step in the technology diffusion and allows the creation of centers of excellence, local training, site visits, regional symposia and educational programs.

While we actively move to further perfect our AI algorithm with more images and additional world-class software developers, we are aggressively moving in parallel towards the next frontier in AI capability.

That is the assessment of patient risk characteristics and the ability to actually predict the likelihood of developing breast cancer.

By incorporating risk characteristics, be they genetic, demographic or breast density scores, we can accurately calculate a patient's risk probability score by comparing a patient's image not to hundreds of thousands of women, as if they all had equal risk characteristics, but to women who possess a similar set of risk characteristics.

This has never been done and certainly never commercialized and available for breast screening.

This moves the screening mammography field beyond age-based screening to risk-adaptive screening.

It makes complete sense to screen high-risk patients more frequently and low-risk patients less so.

We expect to have this capability in the market in 2020, and we anticipate showcasing an early version of this offering at the Radiological Society of North America, RSNA, in November of this year.

Our developers have been working in this area for nearly 2 years, by combining our efforts with lead researchers at The Karolinska Institute in Stockholm, Sweden.

Karolinska is well-known and is one of the world's foremost medical research universities.

We have taken a quantum leap forward by working collaboratively with them, with our agreement.

Together, we possess the world's largest combined database of AI-interpreted images from virtually all OEM vendor imaging equipment providers, along with a huge volume of patients and their associated risk characteristics.

Stacey will outline in further detail this agreement during her section today.

Note, however, that this is not a distant reality that is many years away.

We anticipate having our first-generation risk prediction algorithm in the market in 2020.

It is our view that the ability to predict in a highly individualized manner will usher in a new era of risk-based screening versus our current age-based screening model.

We are at the threshold of personalized disease management as a new approach to breast cancer, where the power of AI software will allow personalized diagnostic care to be delivered in a unique and highly personalized way.

We firmly believe that ongoing collaboration of industry, academia and high-volume clinical sites is essential to creating new categories of AI capability.

Thus, we have formed a medical advisory board of high-profile thought leaders from prestigious sites around the U.S. and outside U.S. locations.

All board members have a unique and lead position on the subject of risk and prediction.

The inaugural meeting of this board will be in early Q2 and we believe will further catalyze our efforts, clinical studies, pre-introduction of offerings in this breakthrough area.

Risk prediction literally changes the conversation between the radiologist and the patient.

We need to work collaboratively with all to shape this conversation.

As we move towards being the innovative pacesetter in clinical AI, we have defined our growth opportunities along 3 clear vectors.

We describe these vectors as horizontal, vertical and diagonal.

The horizontal axis is our current business, mammography.

With 40 million new mammograms each year, we are at the beginning of a new AI adoption cycle.

With a well-timed Q4 financing, we have actively accelerated several mammography AI offerings -- some new offerings, one of which I've already mentioned, our risk prediction module.

Others, now accelerated, include the development -- which would include a number of projects that we are carefully monitoring.

And again, these developments will be showcased at the Radiological Society of North America this coming Q4.

As outstanding as our current AI -- ProFound AI offering is, by expanding sales to new sites, we are rapidly gaining access to thousands of additional images that we can use to have an even better version of ProFound AI released in the market in 2020.

Think of ProFound as a platform, not as a product.

And as a platform, we can look backwards at prior images, find the same area of concern in a singular image and see the history of progression and temporal change in that individual patient image.

No AI breast cancer company has ever commercialized this ability, and it is another crucial module for 2020 commercial release.

We use this ability to look at the kinetics of prior change.

We can then factor in patient risk characteristics to extrapolate and predict cancers within the next 12 to 24 months.

We are not talking about lifetime risk, but the far more important near-term risk.

This ability to look backwards, also see where the patient is today and then project forward and risk predict, is what takes ProFound to a full platform offering.

It is now being vastly accelerated and will be in commercial release in 2020.

The vertical vector is the development of our ProFound cloud offering, which is an initiative that has been developing within the company for more than 18 months and is now moving towards its next level of deployment and testing.

Our recent funding has accelerated this release as well and allowed us to take critical projects that were moving at a more measured pace and aggressively move them forward.

The promotion of Jonathan Go to our Chief Technology Officer is both to honor his achievements with ProFound AI, but to also empower him to move with even greater velocity.

A key driver for ProFound cloud is the reality that there is an array of componentry at radiological sites.

With several image-capturing systems available, a dozen or more possible workstations, at least 15 PACS vendors to transport images and data, each site represents a novel configuration of technology and clinicians.

The ProFound cloud solution simplifies this by allowing AI processing of images in the cloud, along with the ability to send those images to any workstation, computer, even an iPad or an iPhone.

Access to AI can be instantaneous, on demand, and customers only pay for the amount of capability they need, when needed, on a per click basis.

It is a highly repeatable and even a more predictable business model, with all upgrades and new functionalities available in the cloud.

ProFound cloud will significantly reduce servers and other hardware necessities.

The associated reduction in cost and service requirements, as cloud-based solutions and service upgrades are efficiently released, will definitively lower cost, improve margins and allow for savings to be passed on to customers.

We see ProFound cloud as not only a huge enabler but the antidote for many 3D tomosynthesis customers who have [dissipated] adoption and may be more price-sensitive than early adopters.

ProFound cloud, using AI technology, when combined with performance enhancements, will allow for a significant improvement in detection and fewer callbacks.

Both of these save money.

Add to this a more than 50% reduction in reading time.

This is not a marketing claim, but this is our FDA claim received this December.

And together, we are uniquely positioned for success in this next phase of tomosynthesis adoption, where a broader market begins to adopt.

The best answer to price sensitivity is lowering customers' total cost and driving value-added features and functionalities that are measurable.

We can do that with our unique AI capability and with ProFound cloud.

The horizontal vector upon which mammography suites -- or mammography suite of capabilities resides via the ProFound cloud is one vector.

The vertical vector allows ProFound cloud to improve efficiency while offering access anytime, anywhere and in whatever format or usage volume wanted.

This allows us to better serve the 3D tomosynthesis market with customizable solutions.

Now we see 9,000 3D tomosynthesis units in the market today, over 7,000 of which are in the U.S. We've seen the market adopting at a rate of 375 to 400 additional units per quarter.

With ProFound AI, we see a robust opportunity for the already-installed base of over 7,000.

As the U.S. installed base of 7,000 3D tomosynthesis units marches towards the total potential of 18,000 imaging systems in the U.S., we believe we can favorably impact adoption patterns and provide enhanced efficacy and cost efficiencies along the way.

We see ourselves assisting partners in the growth of 3D tomosynthesis, not competing with them.

Finally, when we combine capabilities on the horizontal axis, ProFound platform, with the vertical axis, ProFound cloud, we clearly see a diagonal vector.

This is the vector of being a clinical AI company involved in new diseases and imaging modalities.

We actually have a good start here.

We have a CT colonoscopy offering that is very robust and is FDA-cleared.

As the company pursued a breast cancer focus with AI abilities, opportunities such as this were not a focus in the past.

They will be a focus now.

With a clinical AI focus, the diagonal vector, our thinking and capability broadens to vast and broader market opportunities.

We have experienced building out an entire prostate MRI product and we've done work in other areas as well.

We are now evaluating all of these imaging opportunities and the related disease diagnostic opportunity.

We don't envision having to build out a full commercial infrastructure for these opportunities.

In most cases, there are logical channel partners.

They provide channel access and the image capture hardware.

Effectively, we will do what we do best, and our partners can do what they do best.

We have the ability to build customized offerings for them to bring to market efforts in this area that could catalyze a number of development agreements.

We're actively working in this direction.

While a direct commercial model is necessary in mammography, we believe, particularly to proliferate, educate and demonstrate value, it may not be as necessary in non-breast cancer imaging areas.

As developments unfold in this area, we will continue to communicate material developments as we drive towards being a broader clinical AI provider of technology.

Now since we often get asked questions about partners and partner relationships, I'd like to share our perspective and align those thoughts with the 3 vectors just outlined.

We are committed to an open architecture and being freely accessible AI technology offered for all partners and all clinicians.

This is the reason we are now nonexclusive in all markets.

We believe that, just as with the Internet, best-of-breed applications and solutions always win.

The market is very efficient in choosing winners and losers, and we want to fully optimize our core competency in AI technology to provide class solutions.

We welcome partnering and collaboration as long as all customers have access.

For in the end, it's all about meeting clinical needs and generating better patient outcomes.

With these principles, we believe we have a natural alignment with PACS partners.

They have a unique focus on high access and seamless movement of images and data throughout the entire health care system, be it in local, regional or cross-country.

PACS vendors also deploy cloud-based technologies and span the entire medical imaging ecosystem powered by an open architecture, anytime, anywhere approach.

This national alliance has us working with the majority of PACS vendors.

In fact, we featured ProFound AI at 2 recent trade shows: the RSNA; and the European version, ECR, European Congress of Radiology.

We did show with PACS vendors in their booths and in the booths of PACS vendors that represented over 75% of the market.

We also know that our expanding customer base, which numbers in the thousands, at both U.S. and OUS sites, is very attractive to our PACS partners.

The inevitable expansion of PACS and cloud capability, over time, will lead us towards more of a pay-per-click revenue model.

And again, we are fairly aligned with this emerging revenue model, which is of interest to the PACS vendors and integral to their business model, and in the future, will be with ours.

We will communicate evolution and development in these relationships as we move forward, and certainly, we'll provide transparency as business models evolve.

Whether it be PACS vendors, mammography OEM, distribution partners, which is a big part of our European go-to-market approach, our North Star is to provide clinical AI solutions that meet both current and future customers' needs and align with how and where they want to be able to access AI-powered, personalized medical imaging solutions.

Now before I turn the call over to Scott for an in-depth financial review, I'd like to also highlight some elements of our therapy business.

Importantly, we are increasingly optimistic on the outlook for our IORT business, Xoft, our accelerated 10-minute radiation therapy solution that's used real-time during breast cancer lumpectomy, that's when the treatment is delivered.

Most of you know that I served as CEO of this business when it was a start-up.

Xoft technology is now being utilized in approximately 40 U.S. sites and approximately 50 sites outside the United States.

Some sites take advantage of the portability of the Xoft system and its ability to be used in non-shielded room locations at multiple patient care sites, beyond the 90 sites mentioned.

We are targeting selling more systems to all of these locations as well as our now more rapidly growing Asian market, where IORT has broad appeal.

Given its small footprint, nonradioactive energy source, no shielded room requirement, and again, a 10-minute treatment protocol, it's an ideal product.

It's an ideal product to democratize care and take it to all patients in all areas, particularly in low cost of care sites that are often closer to where patients reside.

Today, with recently published and very robust clinical data, one in particular that tracked over 1,000 IORT-treated patients for 55 months, we are excited about the growth prospects for this business.

With a riveted focus on Xoft, as stated earlier, we have grown therapy product sales in 2018 22% above 2017 revenues.

The data outcome from multiple long-term studies clearly shows recurrence rates that rival that of radiation therapy, a therapy that can take up to 6 or 7 weeks.

With the ability to offer a single treatment during lumpectomy delivered real-time in an unshielded room, Xoft is earning a meaningful place in the spectrum of cancer treatment.

It is also FDA-cleared for all treatment areas and is enormously scaled to areas ranging from rectal to gynecological cancers or even prostate cancer, where we have now recently completed our first IORT treatment following a robotic radical prostatectomy.

From a financial perspective, as Scott will detail, the results of our strategic decision to discontinue a subscription service model to customers in our skin brachytherapy business had a beneficial impact on our overall business in 2019.

We have not left the skin business and offer units for sale into that channel.

We simply are not in the business of placing systems and assuming that dermatologists will drive the volume.

In closing, we remain focused on unlocking the significant inherent value that exists in both iCAD's AI imaging and Xoft's enormously scalable electronic brachytherapy technology.

We are excited, and I would say blessed, to have 2 core technologies and huge opportunities in both AI and therapy.

And we are riveted on the achievement of our near-term quarter-by-quarter financial and operational goals.

We recognize that our long-term sustainable success has to be won and earned quarter-by-quarter and by logically constructing measurable metrics and key performance indicators, KPIs, that drive our success and allow our investors to accurately assess our success and progress on a regular basis.

With that, I'd like to now pass the call over to Scott.

Scott?

Thank you, Mike, and good afternoon, everyone.

I would like to review the company's fourth quarter and full year 2018 financial highlights, which were released earlier this afternoon.

As I will only be discussing the highlights from our financial results, I'd like to refer you to our annual report on Form 10-K for more specifics and details.

The company expects to file its Form 10-K for 2018 with the SEC prior to March 30, 2019.

Before I begin with the financial highlights, I would like to note that during my comments today, I will be referring to certain non-GAAP financial measures.

Management believes that these measures provide meaningful information to investors and reflect the way that we view the operating performance of the company.

You can find a reconciliation of our GAAP to non-GAAP measures at the end of our earnings release.

With that covered, I will now summarize our financial results for the fourth quarter.

Given both the management and strategic changes in our business, I thought it would be helpful to first compare our fourth quarter to the third quarter in 2018 and then follow up with the prior year quarter.

For the fourth quarter of 2018, total revenue was $7 million, which represented an improvement of $0.8 million or 13% over the third quarter 2018.

The detection segment increased by approximately $1 million, with strong growth on both our 2D and 3D businesses.

This increase was offset slightly by a $0.2 million decrease in the therapy business, which represents 2 systems in Q4 as compared to 3 systems in Q3.

Moving on to the fourth quarter of 2018 as compared to the fourth quarter of 2017.

From a revenue standpoint, fiscal year 2018 represented a year of challenges for the company.

On the therapy side of the business, as we announced in January 2018, the company made a strategic decision to exit the skin subscription business.

In addition, detection revenue declined as compared to fiscal year 2017 due primarily to underlying weakness related to sales of our OEM partners' products.

For the fourth quarter of 2018, total revenue decreased by approximately $0.9 million or 12% from $7.9 million in the fourth quarter of 2017 to $7 million in the fourth quarter of 2018.

The overall decrease is due to an approximate $0.3 million or 6% decrease in detection revenue and a $0.6 million or 24% decrease in therapy revenue.

The $0.3 million decrease in detection revenue was due primarily to product revenue, with the decrease in revenue attributable to our OEMs of approximately $600,000, offset by an increase in our direct sales of $300,000.

We believe this is a positive indicator of our decision to increase our sales force in 2019 and continue to pursue the significant installed base of 2D and 3D customers.

On the therapy side of the business, revenue decreased by $600,000, with $0.5 million due to the exit of the skin subscription business and the remainder of the decrease due to the fact that we sold 1 U.S. and 1 OUS system in the fourth quarter of 2018 as compared to 4 OUS systems in 2017.

As I'll also mention further down, the move to exit the skin subscription business had benefits for the company in improved focus on IORT as well as reducing our cash burn and expenses.

Moving on to gross profit.

On a pure dollar basis, gross profit for the fourth quarter of 2018 was $5.4 million as compared to $4.3 million in the fourth quarter of 2017, a $1.1 million quarterly improvement.

On a percentage basis, gross profit for the fourth quarter of 2018 was 78% of revenues as compared to just 55% in the prior year quarter.

This reflects an impressive 23% improvement year-over-year.

Both the gross profit dollar and the gross profit percentage improvement were primarily attributable to the company's exit from the skin subscription business.

On to operating expenses.

Operating expenses for the fourth quarter of 2018 totaled $8.2 million.

This represented a $0.3 million decrease from the fourth quarter of 2017.

It should be noted that the prior year quarter included a $2 million impairment charge related to our therapy business.

Absent this charge, operating expenses increased $1.7 million or 26%.

This year-over-year increase in operating expenses is reflective of fourth quarter legal expenses related to board and management turnover, the debt placement, severance and incremental sales and marketing expenses.

As we look to 2019 and continue to build out the infrastructure necessary for market success of ProFound AI, we expect our operating expenses to trend somewhat higher in 2019 as compared to 2018, especially in the first half of the year.

Now summarizing our profit metrics.

GAAP net loss for the fourth quarter was $3.4 million or $0.20 per share compared to net loss of $4.2 million or $0.26 per share in the comparable prior year quarter.

The $0.8 million improvement in our net loss was, again, due to the absence of the $2 million goodwill impairment charge in the prior year period and the increase in gross margin from the exit of the skin subscription business, offset by increased operating expenses in the fourth quarter.

On a non-GAAP adjusted EBITDA basis, the net loss for the fourth quarter of 2018 was $0.6 million.

This represented a $0.7 million improvement over the prior year period, and the improvement in our loss was primarily driven by the reduced expenses associated with the skin subscription business.

Moving on to the balance sheet.

We ended 2018 with $12.2 million cash on hand.

This compares to the $9.4 million we had on hand as of December 31, 2017.

Q4 2018 includes the approximate $7 million private placement we successfully completed in the fourth quarter of 2018.

In this transaction, we sold unsecured subordinated convertible debentures due 3 years following issuance to certain institutional and accredited investors, including participation by all the directors and executive officers of iCAD.

Interest on the debentures will be paid semiannually in cash at a rate per annum of 5%.

At any time prior to maturity, the debentures are convertible into shares of iCAD common stock at a conversion price of $4 per share, which will be subject to adjustment.

With this financing, we entered 2019 in an extremely strong financial position.

We are focused on preparing for the successful launch of what we believe is a transformational product, ProFound AI.

In addition, our strategic decision to exit the skin subscription business continues to benefit our overall business and we see meaningful growth opportunities within IORT in our therapy segment.

This concludes the financial highlights section of our presentation.

I would like to now turn the call over to Stacey.

Stacey?

Thank you, Scott, and good afternoon, everyone.

Let me begin with some high-level comments on our overall performance in 2018.

While we faced some top line headwinds during the year, as Scott indicated, we achieved significant gross margin expansion and substantial reductions in both operating expenses and cash burn.

In our software business, mammography product revenue declined 7% year-over-year in 2018.

The most significant challenge confronting us here was that business with our largest OEM partner, GE, was relatively weak throughout the year.

This weakness was partially offset by stronger direct sales efforts in 2018.

With all of that said, we headed into 2019 with strong momentum in our business and a lot of confidence in our ability to execute our plans and accelerate growth.

As you know, in December of last year, we received FDA clearance for ProFound AI, a powerful solution that provides our customers with access to the most cutting-edge breast cancer detection software in the marketplace today.

The FDA clearance was based on positive clinical results from a large reader study completed in 2018 that showed increased cancer detection rates, reduced false positive rates and patient recalls and a significant decrease in reading time.

Specifically, ProFound AI has demonstrated an improvement in cancer detection rate by an average of 8%, a reduction in the rate of unnecessary patient recalls by an average of 7% and decreased radiologist reading time by more than 50% on average in clinical testing.

We launched ProFound AI in the U.S. shortly after receiving FDA clearance in late 2018 and even completed a limited number of sales prior to year-end.

The first U.S. ProFound AI sites went live in January, and as Mike indicated, feedback from our initial ProFound AI customers has been highly positive.

We have also identified our initial KOL sites, which will be important in establishing critical external validation for our products from true thought leaders in the space.

We are now actively building key reference sites, testimonials and educational programs designed to accelerate adoption.

We are excited about the large total addressable market opportunity of approximately $500 million over the next 5 years or so for this solution, which we firmly believe has already begun to fundamentally transform breast cancer detection and patient care.

With that said, we are taking a very measured and pragmatic approach to executing this launch.

We have planned for a controlled rollout to ensure that our cross-functional teams are prepared.

Our focus in the first half of 2019 will be on making investments to expand our direct sales team in the U.S. and Europe as well as our distributor relationships in Europe.

For our U.S. direct team, we have already established strong marketing and promotional tools, including an ROI tool for customers, updated collateral materials and website content as well as media placements that highlight ProFound AI.

We've also completed an extensive sales and service training boot camp.

In addition, GE's official launch of ProFound AI is expected to occur very shortly.

Training for their U.S. sales team was completed in February.

Importantly, however, I'd like to reiterate what Mike stated earlier.

As we continue to make progress towards moving to the cloud with ProFound AI and building out the ProFound platform with our increasingly robust commercial resources, our dependence on imaging OEMs overall will continue to decline as customers will have ready access to our best-of-breed AI solutions from many channels and partners.

We continue to be open to nonexclusive relationships as our goal is to promote awareness and access to our revolutionary cancer detection solutions today as well as in the future when we move into risk prediction.

While we are focused on executing a highly successful rollout of ProFound AI, we also continue to invest in future product development.

Our ProFound AI product for 2D mammography is currently pending CE Mark in Europe.

We previewed this product at the recently held 2019 European Congress of Radiology, or ECR, the second-largest radiological meeting in the world.

Also at ECR 2019, we showcased ProFound AI for 3D mammography in our booth and led 2 clinical presentations.

In addition, we hosted a very well-attended customer event, where clinical leaders discussed their perspectives on how iCAD solutions are transforming breast cancer detection, shared their personal experiences with ProFound AI and the research and the clinical practice and detailed their vision for breast cancer risk assessment.

As Mike discussed, risk assessment is the next critical area of focus for iCAD.

Our recently announced partnership with researchers from The Karolinska Institute builds on an existing research agreement whereby these researchers developed a breast cancer risk prediction model using information identified in mammography images provided by iCAD's AI cancer detection and breast density assessment solutions.

Promising early results based on mammography images from over 70,000 Swedish women enrolled in the Karolinska mammography project for risk prediction of breast cancer, or Karma study, were published in Breast Cancer Research in 2017.

These data indicated that the model developed enabled early identification of women who were at high risk for breast cancer and it was determined additional examinations were warranted.

Since its publication, these results have been improved upon through the use of iCAD's latest, latest ProFound AI algorithm.

Among other things, the model now takes asymmetry of mammographic features and masking of tumors due to dense breast tissue into consideration.

iCAD and The Karolinska Institute researchers now intend to collaborate to develop an innovative solution for commercial use to assess an individual's risk of developing breast cancer.

There's no other product on the market today that combines this level of imaging data with other individual characteristics, resulting in a tremendous improvement in accuracy and a game-changing shift from what is now age-based screening to a risk-adaptive screening paradigm.

On the therapy side, product sales in this segment increased 22% in full year 2018 as compared to last year.

This growth was primarily driven by continued strong performance in our Breast IORT business.

We sold 2 IORT systems in the fourth quarter, 1 in the U.S. and the other in Spain.

We also closed 4 rental agreements in Spain.

With these additional Spanish centers, we now have 7 installations across 4 major regions of Spain.

We continue to be pleased with our growing installed base, which currently includes approximately 40 U.S. centers and close to 50 OUS centers, treating patients with IORT and/or gynecological applicators.

In addition, we saw nice growth in utilization, with a 7% increase globally in full year of 2018 Balloon applicator sales versus 2017.

Increased interest in the U.S. continues to be driven by growing favorable clinical data, most recently, positive results from a series of 1,000 tumors treated with IORT therapy using the dock system for early-stage breast cancer in a single site trial, were published in the Annals of Surgical Oncology.

This prospective trial enrolled a total of 984 patients with 1,000 early-stage breast cancers from June 2010 to August 2017, with a median follow-up of approximately 3 years.

Study results show that breast cancer recurrence rates of patients who were treated with IORT using the dock system were comparable to those seen in the TARGIT-A and ELIOT studies, which evaluated IORTs in different technologies.

Our OUS momentum is being largely driven by progress in key markets such as China, Taiwan, Spain, India and Australia.

In addition, we are seeing increasing interest in Japan, a large target market for us in the future.

We also have multiple active projects underway to commercialize new applicators and open up additional market opportunities.

We are specifically experiencing growing global interest in general IORT applications, including prostate, brain and rectal cancer, as a result of key clinical research being conducted globally.

As you can see, we are experiencing a number of positive trends in our IORT business.

In fact, we view IORT as the most significant long-term growth driver in our therapy segment.

As such, we are seeking a commercial partner to lead the sales and marketing efforts for our skin brachytherapy business.

There's no specific time line for completion of this process, but we are confident in our ability to identify a leader in the dermatology space as a commercial partner.

In summary, we made significant progress on multiple fronts in the fourth quarter, and we continue to make key targeted investments in our organization that we believe will accelerate growth in 2019 and create a sustainable leadership position for iCAD for years to come.

I would now like to turn the call back to Mike for additional comments.

Mike?

Thank you, Stacey.

Before opening up for questions and answers, I wanted to make a few qualitative comments about what we expect in the first half of 2019.

As I highlighted in my earlier section, we are using our enhanced cash position in 2 key areas: one is to build commercial infrastructure for our robust launch of ProFound AI; the second is for the development of core programs to build out the ProFound platform, our horizontal axis of mammography, and the acceleration of ProFound cloud, the vertical axis.

Specific to our mammography business, the largest revenue and margin driver for 2018, we now have the commercialization infrastructure in place.

We are working to synchronize what I call 3 cycles or clocks.

The first of these is a 90- to 120-day cycle or clock, is the scale of time for recently hired sales reps.

Some were added in Q4, and we now have the full sales service infrastructure in place as we end Q1.

The second cycle or clock is the ramp-up of accounts in our pipeline and funnel.

Although we pre-announced ProFound AI in late Q4, we cannot begin to quote prices and hit budgets or sales cycles until the very end of 2018.

We see a 90- to 120-day cycle process, where -- from the time of first presentation to account, to the close of a sale.

We also have a third cycle or clock, which involves site inspection, installation, connectivity to the specific reading environment, service and full training.

This too is a 90- to 120-day clock or cycle.

These cycles or clocks, though somewhat lengthy, are not sequential.

Proper and meticulous execution will condense what on the surface would appear to be a 270- to 360-day cycle, if it was all added sequentially, and synchronize them to a less than 180-day full cycle.

In certain occasions, we hit full cycle sooner, thus our initial sales as reported by a press release last week.

Considering the above, I am reiterating what we've been communicating since I joined the company in Q4.

We see expense build in Q1 and Q2 and revenue that will begin to move meaningfully upwards and show increasingly greater traction in our second half, Q3 and Q4.

We traditionally do have a seasonal dip in our business, from the fourth quarter to the first quarter of the following year.

Much of this relates to hospital purchase cycles and fourth quarter surges.

We will offset some of the seasonal effect with the recent flurry of early-adopting ProFound AI sites.

Early adopters tend to move quickly, but represent a smaller percentage of the market.

Additionally, while we have current customers teed up for purchase, waiting for approval or approved but not installed, we have other sites that are not renewing or purchasing service contracts on older versions of CAD, nor will they upgrade to a better version of our digital offering, our prior digital offering, while they are in queue or teed up for ProFound sales and moving on to the ProFound platform.

In sum, we expect to have some year-over-year slippage in service and upgrade revenues while ProFound AI begins to scale and install.

Also, since we won't have full cycle or clocks fully synchronized until this summer, in certain cases, service install and connectivity may lag actual product procurement and revenue realization.

In the end, in 2019, we do anticipate year-over-year improvement, though with higher expenses in the first half.

We expect sequential quarter-over-quarter sales improvement and therefore enhanced operating leverage in the second half of the year as expenses stabilize and revenues scale.

We see a partial quarter of added expenses in Q1 and the full expense of our infrastructure and development cost in Q2 as our resource additions are fully reflected and then begin to stabilize at the end of Q2.

The install of ProFound platform also allows each of our 2019 installed sites to be an outpost and virtual amplifier, as new offerings such as risk prediction licenses arrive in 2020.

Obviously, this expansion will lead to enhanced ASP of purchases when introduced in 2020.

In sum, we continue with our outlook for both the first and second half as presented in Q4 2018, but we are executing a well-controlled launch with precision and with full synchronization of the above-stated cycle.

And with that, I'll turn it back to the moderator and questions and answers.

(Operator Instructions) Our first question comes from François Brisebois, Laidlaw.

There's a lot in that call to digest here.

But I was just wondering, in terms of the AI consultants, can you talk a little bit how you vet them?

This is so new here, you guys' approach, I'm just wondering how these consultants or reps, how you find these people?

Thanks, Frank.

That's a good question.

We spend a lot of time vetting salespeople, because we're looking for a very specific type of sales rep, and they do indeed come from the mammography world, with experience.

We see that there are basically 3 levers in a sale.

There's a clinical sale, which you've got to get past.

There's a technical, what you might call operational sale, where you have to understand the complexity of the environment.

So you've got clinical and I'd say, technical, and then there's an economic sale, which is up there with what I call the mahogany -- not mahogany, [polished] and carpeted suite.

We need sales reps that can actually move up and down this pyramid.

That can actually have clinical expertise, that are technically astute, understand connectivity and are economically sanguine and can talk about return on investment and how the cost of a license can be offset by the dramatic savings and workflow as well as the performance improvement.

So we're really looking for people that can move up and down this pyramid.

And we look very hard, we probably interview 10, 15 salespeople for every 1 that we ultimately select.

Okay, great.

That's very helpful.

And then I'm wondering, on the IORT, you guys get all that data now.

In terms of growth, I'm just wondering, is there any potential for patient advocacy or whatnot?

Just because it seems like the data is kind of there.

I'm wondering when you guys think that, that will start kicking into the top line a little more?

Again, a great question, and I do think that this is an area that some people call one of the best-keep secrets in radiation therapy and that we do have a product that treats in a non-shielded room, could be near patients and you could do it during lumpectomy.

So I would say, the answer to your question is yes, and that the more we start amplifying the story, let people know of its availability, start getting ready for times like Breast Cancer Awareness Month to let people know that this is an option that is available, and particularly, with the clinical data that we now have, we've earned the right to be able to claim that we have equivalent recurrence rates to 6 or 7 weeks of therapy.

I could tell you that in the early days of CAD, one of the big drivers for adoption was working with Susan G. Komen and Why Me and a number of women advocacy organizations.

And it's certainly on our radar screen to be able to do similar things with this offering to get the word out.

The more people know, the more that will drive sales.

Okay, great.

And if I can just have one quick one here.

You guys just had ECR.

Can you just compare ECR versus RSNA and the response you had and just the thought process a little bit?

Stacey mentioned you presented a 2D machine.

How does that -- how do you make sure that 2D mammography doesn't kind of cannibalize any potential to move up right away to the 3D?

Why don't I make a comment about 2D and 3D, and Stacey can talk about the show.

What is interesting is that with the somewhere between 8,000 and 9,000 systems in Europe, that there have been about, from our data, about 1,500 of those that have moved to tomo.

There are a lot of systems that are still -- a lot of sites that are moving from analog, film still, to digital, so almost first-generation analog into digital and then into tomo.

In certain cases, they move from analog all the way to 3D tomo.

So therefore, when it comes to Europe, we see a market for 2D that's as robust as the market for 3D.

In fact, for many of the sites, they use 3D for definitive diagnosis and they use the digital for screening.

But in some cases, they are moving directly from analog to 3D.

The reason for the ProFound 2D launch, which was a big part of the -- of our ECR, was for this rather substantial, several thousand sites, that are either in analog going to digital or maybe on a first-generation digital that we want to upgrade.

Stacey, you want to talk about the other parts of the ECR?

Yes, I'll just add to what Mike was saying.

The market in Europe, Frank, is very different than the U.S. The 2D market in Europe is still growing and is expected to continue to grow for at least the next 5 years.

And there are about -- if you look at just the key countries alone, so U.K., Italy, France, Spain, top 5 or 6 European countries, there are about 6,000 2D machines in those countries, and only about 10% of them have CAD software attached to them.

So we have a tremendous opportunity with our ProFound AI product for 2D to go back into that large installed base and sell them a software product and then also to go after the new 2D systems that are still being sold.

So it's a different market dynamic, and we actually expect a fair amount of our growth in Europe to come from sales of the 2D product, in addition to 3D.

In terms of the ECR, we really had a tremendous presence at that show this year.

We had a larger footprint, a lot more visibility in terms of presentations.

We had oversubscribed customer events, with risk assessment being one of the key topics.

And we really have invested quite a bit more in the EU this year than we ever had in the past.

So we now have 5 KOL sites up and running, so it's cut across all the major imaging vendors.

So we have sites with GE, with Hologic, with Siemens.

And we recently have put in place 4 new distributors in Europe that collectively cover about 8 or 9 countries.

So in addition to that, we also added a direct sales rep in France.

France is actually the largest mammography market in Europe, and we thought that it could command a direct presence.

But we really have put a lot more focus on the European market than we have in the past, and we expect that to pay off with increased growth in that geography in 2019.

Our next question comes from Per Ostlund, Craig-Hallum Capital.

Actually, since I'm going to follow up or dovetail off of Frank's question right away, just since it was on my list as well.

When you look at that 2D market in Europe, Stacey, and you mentioned or you alluded to the fact that I think CAD is still very, very underrepresented with 2D, what do you see as the gating factors to adopting CAD and how does ProFound AI, do you think, accelerate the adoption of CAD within the 2D market there?

Yes, it's a great question, Per.

So historically, what has held back adoption in Europe is that it's a very different reading paradigm than in the United States.

So today, and historically, every mammogram is read by 2 radiologists in Europe.

And so the challenge for us, and we have already started to prove this in some clinical studies, is that you have to prove that 1 radiologist, plus CAD, is better than 2 radiologists, right?

And then you can obviously show an associated economic impact, favorable economic impact, from that.

So that's the first thing.

The second thing is that the increased performance that comes from the current-generation AI technology is much greater than the previous CAD products, right?

So it's actually much easier to show the clinical value proposition, along with the economic value proposition, than it was in the past.

So combination of kind of 2 things there that are at work.

If you look at the major trade shows, and we've actually started to invest a fair amount more in some of the in-country trade shows, whereas in the past, you didn't see a lot of podium presentations on CAD.

That whole dynamic has changed and now there are a lot of presentations going on about CAD, a lot of discussion among the KOLs.

And it's just a situation where Europe has lagged the United States in terms of that adoption, much like the 3D adoption is behind the U.S. by probably 3 or 4 years.

So it's kind of a convergence of a number of factors, but we're definitely seeing increased interest and increased growth in terms of 2D and ultimately, 3D as well, but the market will still be much bigger for 2D for quite some time.

Sure, okay.

That makes sense.

Coming back to your release late last week about ProFound AI in the U.S., and so the number of sites that are already using the tool across the country, just wondering, it's very early, but I'm wondering if you can characterize a little bit of some differing mix factors, maybe in terms of which mammography OEMs are involved, is it kind of spread amongst different ones, or is it still somewhat concentrated?

How have these customers come to ProFound?

Or is it via your direct sales?

Is it through OEMs, or a little bit of both?

And then is this a situation where these are sites that were already tomo sites that are now ProFound customers?

Or did ProFound actually catalyze that upgrade to tomo at the same time?

Yes, it's a great question and lots of stuff in there, Per.

Let me see if I can remember each of the different questions.

So what I would say first off, is that the early installations are a mix of GE and Hologic sites.

Some of the GE sites were using the version 1 tomo that was exclusive to GE and now have upgraded to the new ProFound AI.

In terms of profile of the sites, they are a mix between leading academic institutions and small community hospitals and everything in between, a mix of hospitals and imaging centers, and they are geographically spread out across the country.

So in all major parts of the country, we have installed successfully a ProFound AI.

And the early feedback, as we mentioned, is very positive.

And the other thing I'll mention is that we also have successfully connected to a large number of different reading environments, so different PACS systems, different independent mammography review stations from the imaging vendors.

So there are -- it's not just one sort of particular reading environment, it's a very large mix of different reading environments.

And that's great for us as we go forward because after we get the first one nailed down and really understand how to maximize workflow with a particular PACS company or imaging review workstation, that makes the ones that come after that a lot easier.

So let's see, did I hit the majority of the questions you asked?

You did, you did.

Well, I guess maybe the one would be how much of this has come via your direct sales force going out versus in tandem?

Okay.

They're primarily direct sales force orders that we have here.

So now, we've now taken our direct sales force from 6 to 11 reps, and these early installations have been primarily sold through the direct sales team.

And I would just add, Per, that we really have 3 value propositions, and some of the mix in each account, in terms of what they value, is different.

There's obviously sensitivity, some people value the sensitivity -- increased sensitivity.

Some people are really interested in the false positive reduction, because fewer callbacks, fewer cost for that facility.

And then there are others that they're all about workflow.

They double, they triple their work time, and they're interested in workflow.

Obviously, we're very successful in sites where we triangulate all 3 of them.

But we find that when those -- when you have 2 or 3 of those things aligned, we wind up with a site that moves from 0 to 60 miles an hour very, very quickly.

Sure.

No, all that makes sense.

And Mike, I -- that actually leads into, I think perfectly leads into a question.

I think the clinical characteristics, the workflow characteristics, I think those have been well-established at this point.

And you referred to a conversion rate of kind of 375 to 400 instruments a quarter, and there are still 5 digits' worth of instruments out there to be converted in the U.S. Given that workflow probably in particular might have been the biggest pushback or among the pushbacks for 3D tomo, is there an opportunity to see that 375 to 400 kind of conversion rate actually now accelerate with a tool like ProFound on the market?

Well, yes.

I believe that the 375 to 400 is actually moving from the early adopter phase, where people are maybe a little more agnostic on -- they want to be the pioneers and may be a little more agnostic on the price side.

And I think as people become more price-sensitive, and I do think we're in that kind of early majority stage now, that we help with those efficiencies.

We help add that value.

We help reduce the time to read the images.

And a lot of these want us to quantify what could a reduction in false positives mean in terms of workups and callbacks.

So I think it definitely makes a big difference there.

And also, just to double up on one of the points that Stacey made, we initially wanted to go to 15 sales reps this quarter, but we decided that 11 sales reps, plus more people on the third clock, the connectivity, the site inspection clock, was actually more potent than going with 15 reps because we were getting over our skis.

We had more sales opportunities being generated and not enough of the second -- the third clock in terms of connectivity.

So we've now put in place a sales infrastructure with support infrastructure that can move more swiftly, even though there are fewer reps than initially planned.

We could certainly add those additional reps, but for the moment, to keep up with these 375 to 400, as well as the already installed base, having a blend of sales and service support people is what's allowing us to synchronize the clock.

That makes a lot of sense.

All right, one more for me.

I wanted to ask about the collaboration with Karolinska.

And obviously, it is early here.

It sounds like a very, very intriguing opportunity to, I guess, marry your imaging with their kind of more demographic, patient-specific type data.

Do you have a sense at this point as to what the regulatory pathway looks like for that offering?

Yes, I would say that we are working with more than one regulatory consultant and the -- where the FDA comes in on this is where you are, or where we are, on what might be a continuum of being predictive versus prescriptive.

The closer you get to being prescriptive, where you literally are saying, "This is this person's risk and therefore, you should do that," that's where the FDA comes in, because you are actually driving practice in a certain direction.

Where you are simply presenting an algorithmically-derived prediction based on data and if you stay at that end of the spectrum, that's where the FDA is a little more sanguine to opportunity.

So we need to walk this path between predictive and prescriptive, and in many ways, allow -- always keep the physician in the middle making the final choice.

Because at the end, what the FDA really gets concerned about, and it was the same was the case when detection was a PMA, was is it going to be diagnostic versus detection?

Is it going to take -- is it going to drive physician behavior in such a way that could be problematic?

So, so far, the early indications are that we're in a good position if we stay on the predictive side and don't cross over into prescriptive.

(Operator Instructions) Ladies and gentlemen, we have reached the end of the question-and-answer session.

And I would like to turn the call back over to Michael Klein for closing remarks.

Well, thank you.

And I would like to close by reiterating some priorities going forward.

To maximize the commercial success with ProFound AI for digital breast tomosynthesis globally, our commercial launch is underway.

We are excited about the compelling growth prospects of this product, particularly in the second half of this year.

We are, number two, accelerating development in 2019 that will extend our lead in 2020 in both our AI business and our very scalable IORT platform with Xoft, using electronic brachytherapy.

Specifically, the above will take form of our expansion from detection into risk prediction on the AI side and the continued growth of our Breast IORT offering as well as non-breast applications in areas such as gynecology, rectal, brain and other disease areas.

And finally, we will continue our transformation into a broader clinical AI company, with the exploration and advance of new AI offerings across radiology areas, imaging modalities and disease space.

We will keep our eyes on new horizons with the recognition that we will thoughtfully manage financials, our controlled rollout of products and ensuring we predictably hit key metrics and performance indicators, all while continuing to articulate with transparency our processes, milestones and successes.

So I want to thank you all again for joining our call today, and we look forward to providing additional updates as the year progresses.

Thank you for joining us.

This concludes today's conference.

You may disconnect your lines at this time.

Thank you for your participation.