Investment Logic in Medical AI: Opportunities in Breast Cancer Screening Driven by Ultrasound-Based CAD and ABVS Technology

The healthcare sector has long been perceived as conservative and insular, yet investment trends often chase and favor hotspots. Currently, healthcare investment is a focal point within the investment community, and trends within the healthcare field are naturally regarded as the most prominent of these hotspots, making it entirely logical that they attract widespread attention. How to pursue hotspot investments within an industry characterized by conservatism is an intriguing topic. This article will begin with an analysis of the industry’s inherent attributes to explore underlying patterns and identify investment opportunities.

The application of artificial intelligence in healthcare has a long history, with commercialization already well established abroad.

Artificial intelligence was proposed as a technical term in the summer of 1956. Over the past five decades, it has made significant progress and become a broadCrossand cutting-edge science.

In the medical field, IBM Watson, a name we are very familiar with, has been serving the MD Anderson Cancer Center for quite some time. By leveraging Watson’s cognitive computing capabilities to extract information from patient records and extensive research databases, it provides clinicians with valuable insights, thereby helping healthcare professionals identify the most effective treatment plans. AI in healthcare also includes companies such as Enlitic, Atomwise, The Human Diagnosis Project, Butterfly Network, and Arterys. However, the unique nature of the healthcare sector dictates that any product must undergo a rigorous approval process before commercial application. In reality, most of these well-known products remain distant from widespread clinical use. We look forward to their early approval and integration into our daily lives.

Currently, the application of artificial intelligence in the medical field is not yet approved for therapeutic purposes, whereas its use in diagnosis has a 40-year history, known as computer-aided diagnosis (CAD). At present, the term CAD technology primarily refers to computer-aided techniques based on medical imaging. Since the 1990s,Artificial Neural Network(Artificial Neural Network, ANN) The rapid development of ANNs has enabled qualitative breakthroughs in computer-aided diagnosis (CAD) within the field of medical imaging. As a mathematical approach that mimics the working principles of human brain neurons, ANNs possess capabilities such as self-learning, memory, and prediction of event progression, thereby facilitating auxiliary diagnosis. In terms of classification and diagnosis, artificial neural network methods demonstrate superior performance compared to traditional approaches (such as probabilistic statistics and mathematical models). Currently, most CAD research remains limited toBreast...and nodular lesions in the chest and lungs; therefore, CAD research on breast and pulmonary nodules can essentially represent the current state of CAD.InMedical ImaginginState-of-the-art levels and current status. Commercial applications abroad are concentrated in these two fields. For example, R2 Technology, founded in Silicon Valley in 1994, developed the world’s first computer-aided diagnosis (CAD) system for mammography to receive FDA approval (in 1997), with cumulative sales exceeding US$1 billion (the company was later acquired by Hologic).

Therefore, CAD is the earliest and most representative application of artificial intelligence in the medical field, and its application in the diagnosis of breast diseases has achieved large-scale commercialization.

The Commercial Application of Overseas Breast CAD Has Not Affected Our Lives

Taking breast diseases, the most mature field in the commercialization of Computer-Aided Diagnosis (CAD), as an example, methods for breast cancer screening include breast self-examination, clinical breast examination (CBE), ultrasound, mammography, tumor marker testing, and genetic sequencing diagnosis. Due to individual differences and a lack of professional knowledge, breast self-examination has a low detection rate and is not suitable for population-wide screening; it serves merely as an auxiliary tool for women to monitor their own health. The main tumor markers for breast cancer include the mucin-like glycoprotein MUC-1 family (such as CA15-3), carcinoembryonic antigen (CEA), the oncogene HER-2 (or CerbB-2), lactalbumin, glycolytic enzymes, cytokeratins, and tissue polypeptide-specific antigen (TPS). In clinical practice, tumor markers are primarily used to detect and prevent metastasis, recurrence, and to assess prognosis in breast cancer. However, due to their lack of sensitivity and specificity for early diagnosis, tumor markers currently hold limited significance in the early detection of breast cancer. Early screening for breast cancer using genetic sequencing is not yet mature. Furthermore, the complexity of tumorigenesis means that our current understanding of genetics cannot fully explain the entire process of cancer development. Therefore, genetic sequencing will not replace imaging-based screening for a considerable period. Although the application of genetic sequencing in breast cancer screening may become feasible as the technology advances, traditional screening methods are expected to remain dominant over the next 10–15 years.

Current screening methods primarily include clinical breast examination (CBE), ultrasound, and mammography. In China, population-based screening programs at the current stage, such as the “Two Cancers” Screening Project initiated by the Ministry of Health, generally adopt ultrasound as the primary modality. This choice is based on the fact that ultrasound is more cost-effective and safer due to the absence of ionizing radiation compared to mammography. Additionally, because dense breast tissue is prevalent among Chinese women, mammography is associated with an unacceptably high rate of missed detections.

Comparison of the Basic Situation of Breast Cancer in China and Abroad

Data source: 2015 China Cancer Statistics

The breast cancer screening market is enormous, and ultrasound is more suitable for use in China, which distinguishes it from the situation abroad. Based on this reasoning, ultrasound-based computer-aided detection (CAD) for breast cancer is the application most aligned with China’s national conditions, whereas no such ultrasound-based breast CAD systems currently exist internationally.

Anticipating the Arrival of the Commercial Era for CAD: AI in Healthcare Is Set to Transform Our Lives

CAD is fundamentally designed to liberate human labor. However, traditional ultrasound equipment relies on manual examinations performed by specialized sonographers, resulting in non-standardized ultrasound images that fail to provide an adequate foundation for CAD applications. Does this mean we have reached a dead end? We have resolved the issue of automated image interpretation at the back end, only to encounter a bottleneck at the front-end stage of image acquisition.

How to Standardize Ultrasound Images: This requires a cutting-edge technology—the Automated Breast Volume Scanner (ABVS). The ABVS is a novel ultrasound diagnostic device that covers the entire breast and does not rely on operator experience, allowing the entire procedure to be performed by nurses or technicians. Current studies have shown that automated breast volume ultrasound scanning yields superior results compared to traditional handheld ultrasound, demonstrating better performance in detection rates and the identification of small lesions.

Let us imagine: ABVS + CAD is essentially a breast cancer screening solution in which the computer-aided diagnosis system provides diagnostic opinions without requiring the involvement of professional ultrasound physicians during the examination process. Having identified this ultimate solution, we can now seek out ultrasound CAD projects based on ABVS following this logic.

Review of this case:

AI in Healthcare — CAD (specifically imaging-based) — Mammography CAD (commercialized abroad with mature business models) — National conditions necessitate breast ultrasound CAD (identifying opportunities) — Non-standardized ultrasound images (a key challenge) — Automated Breast Volume Scanner (ABVS) enables standardization and does not require specialized sonographers — ABVS + Ultrasound CAD (the solution, pinpointing investment directions).

Key Takeaways from This Case:

1. Trending hotspots can also emerge in the traditionally conservative healthcare industry

2. Understanding the historical evolution and overseas development process can help avoid a fragmented perspective and enable a comprehensive overview.

3. Only large-scale commercial application constitutes a true business model and represents the actual investment target.

4. National conditions breed opportunities; analyzing differences means clarifying investment directions

5. Mastery of professional domain knowledge and information can yield unexpected insights

Author: Liu Yun, Huayi Capital