Breast Cancer and AI

"Early detection and accurate diagnosis are crucial for successful breast cancer treatment. AI is emerging as a powerful tool in the fight against breast cancer, offering new hope for improved patient outcomes."

Breast cancer remains a significant global health challenge, being the leading cause of cancer-related deaths among women. Early detection and personalized treatment are crucial for improving outcomes, yet issues like missed diagnoses and access disparities persist despite advancements in screening. Artificial intelligence has recently emerged as a transformative tool in breast cancer care, enhancing diagnostic accuracy and personalizing treatment.

In this week’s edition of AI Horizons, I will explore the intersection of breast cancer and AI, discussing groundbreaking research, innovative tools, and the promise this collaboration holds for the future.

The Silent Battle: Breast Cancer's Global Impact

Breast cancer remains the most prevalent cancer among women globally, significantly impacting individuals, families, and healthcare systems. According to the World Health Organization, in 2022 alone, 2.3 million women were diagnosed, and 670,000 succumbed to the disease, highlighting its immense global burden. In the United Kingdom, breast cancer is the most commonly diagnosed cancer, with approximately 56,000 new cases annually. It predominantly affects women, with 1 in 7 women developing breast cancer during their lifetime. While the disease can occur at any age, the majority of cases in the UK are diagnosed in women over 50. Demographically, breast cancer disproportionately affects women, although men account for less than 1% of cases. Risk factors include age, genetics, hormone replacement therapy, obesity, and alcohol consumption.

Efforts to reduce the global burden of breast cancer focus on awareness campaigns, early screening programs, and improved access to advanced treatments. In the UK, initiatives like the NHS Breast Screening Program have significantly enhanced early detection, contributing to better survival rates. However, traditional screening methods, primarily mammography, are not infallible. Studies indicate that approximately 20% of breast cancers are missed during screenings, underscoring the urgent need for enhanced diagnostic tools and algorithms.

The Advent of AI in Breast Cancer Screening:

Artificial intelligence (AI) is no longer a futuristic concept; it's revolutionizing healthcare today, particularly breast cancer screening. As a surgeon and futurist, I’ve been intrigued by the advancements in this field, especially the potential of AI to transform patient care.

AI's impact extends beyond diagnostics, reshaping how we approach patient care. Tools like PathAI and Qure.ai analyze imaging, pathology, and genomics to guide targeted therapies or suggest follow-up interventions, improving care coordination. Breast cancer screening is notoriously demanding, but AI lightens the load by automating repetitive tasks, allowing radiologists to focus on complex cases. Moreover, by ensuring consistent diagnostic quality, tools like Mia can make high-quality care accessible even in underserved regions.

AI is becoming a beacon of hope in the fight against breast cancer. It improves diagnostic accuracy, reduces false positives, detects cancers earlier, and personalizes screening strategies based on a patient's unique risk profile. Computer-aided detection tools streamline workflows, enabling radiologists to provide more efficient care.

AI isn’t just enhancing breast cancer detection; it’s transforming lives—offering more accuracy, equity, and hope for patients worldwide. As I witness these advancements, I can’t help but feel optimistic about the future of healthcare and the lives we stand to save.

Mia by Kheiron Medical: Setting a Benchmark in Breast Screening AI

Mia® (Mammography Intelligent Assessment) is a breakthrough AI platform by Kheiron Medical Technologies for breast screening. Trained on millions of mammograms, Mia has been designed to support radiologists by acting as an independent second reader during screenings. The platform includes several solutions such as Mia® Reader (assists radiologists in making critical breast-screening decisions), Mia® Triage (flags suspicious cases for more accurate oversight), Mia IQ™ (an AI-enabled solution for automating image quality control), and RSViP™ (an AI-enabled scheduling tool that helps manage backlogs and prioritise women who need breast screening the most).

Mia® Triage supports radiologists by automating up to 25% of the reading workflow, addressing workforce shortages, fatigue, and high image volumes. The AI system aims to improve productivity and accuracy, reduce missed cancers, and lower false positives. It also seeks to build patient trust by providing reliable cancer detection, reducing unnecessary biopsies, and minimising recall anxiety. In a prospective study involving over 25,000 women in Hungary, Mia improved cancer detection rates by 7%, with 83% of the additional cancers detected being invasive. Another trial in the UK reported a 12% increase in detection rates and up to 30% workload reduction for radiologists.

Mia™ is currently undergoing multiple deployments and clinical studies across multiple hospital sites in the UK to ensure its safety and effectiveness. As a surgeon, I see its potential in reforming breast cancer detection and treatment in the future.

Recent Advances in AI and Breast Cancer: Highlighting Key Research Papers

A recently published research paper titled "Nationwide real-world implementation of AI for cancer detection in population-based mammography screening" explores the impact of AI on breast cancer detection in Germany's mammography screening program. The study, PRAIM, involved 463,094 women aged 50-69 years and compared AI-supported double reading to standard double reading by radiologists. The results were quite remarkable. The AI-supported group had a breast cancer detection rate of 6.7 per 1,000 women, which was 17.6% higher than the control group's rate of 5.7 per 1,000. This means that AI helped detect more cases of breast cancer without increasing the recall rate, which, in fact, was slightly lower at 37.4 per 1,000 in the AI group compared to 38.3 per 1,000 in the control group. Moreover, the positive predictive value (PPV) of recall was higher in the AI group (17.9%) compared to the control group (14.9%). The PPV of biopsy was also higher in the AI group (64.5%) versus the control group (59.2%).

This study suggests that integrating AI into mammography screening can significantly improve breast cancer detection rates without negatively affecting recall rates. It also highlights the potential of AI to reduce radiologists' workload, making the screening process more efficient and accurate.

Another article titled "Artificial Intelligence in Breast Cancer Diagnosis and Treatment: Advances in Imaging, Pathology, and Personalized Care" explores the transformative role of AI in breast cancer management. It highlights AI's significant advancements in imaging, pathology, and personalized treatment, improving diagnostic accuracy and patient outcomes. AI enhances mammography, MRI, and ultrasound imaging, rivalling expert radiologists in accuracy. In pathology, AI improves biomarker detection, aiding in precise diagnosis and treatment planning. Personalized medicine benefits from AI's predictive power, helping predict risk stratification and treatment response. The article also discusses the challenges of AI integration, including data variability, ethical concerns, and real-world validation. Despite these challenges, AI holds significant potential to revolutionize breast cancer diagnosis, prognosis, and treatment.

CHIEF (Clinical Histopathology Imaging Evaluation Foundation) is another groundbreaking AI model developed by scientists at Harvard Medical School. This versatile AI system, akin to ChatGPT, is designed to perform various diagnostic tasks across 19 different cancer types, including breast cancer. CHIEF stands out because it doesn't just focus on one aspect of cancer diagnosis. It can detect cancer cells, predict a tumour's molecular profile, and even accurately forecast patient survival. Trained on millions of images, CHIEF interprets both specific sections and whole images, providing a holistic view of the tumour.

In studies, CHIEF achieved nearly 94% accuracy in cancer detection and outperformed current AI methods across multiple datasets. By identifying patients who may benefit from experimental treatments early on, CHIEF can revolutionize cancer diagnosis and treatment, making it more efficient and accessible worldwide. In essence, CHIEF represents a significant leap forward in the fight against cancer, promising to enhance clinicians' ability to evaluate and treat cancers more effectively.

Recently, MIT researchers have developed an advanced AI model called “Mirai” that can detect breast cancer up to five years before a clinical diagnosis. Mirai uses deep learning algorithms to analyze mammography images and accurately predict precancerous changes in breast tissue. It stands out for its ability to identify invasive cancer cells in normal tissue, offering the potential for earlier and more effective treatments.

Mirai's performance has been tested on screening data from seven hospitals in five countries, and it has shown better results than traditional risk models. It has the potential to improve disparities in breast cancer outcomes by providing equitable and reliable risk assessments for diverse patient populations. This breakthrough could significantly improve survival rates, especially for women at high risk of developing breast cancer.

The Road Ahead: Integrating AI into Clinical Practice

Integrating AI into clinical practice for breast cancer detection holds immense promise but requires careful consideration. Regulatory approval ensures AI tools meet stringent safety and efficacy standards. Training and education are crucial to equip radiologists and healthcare professionals with the skills to use AI tools effectively. Continuous monitoring systems must be implemented to assess AI performance and address any biases or inaccuracies over time. Building patient trust is vital, emphasizing that AI tools augment rather than replace human expertise. Although AI in breast cancer care is still in its early stages, ongoing advancements suggest a bright future. Future models will incorporate imaging, pathology, genomics, and patient history data for comprehensive decision-making. AI-powered wearable devices may enable continuous monitoring of breast cancer survivors, ensuring timely detection of recurrences. Scalable, cost-effective AI solutions will make advanced breast cancer care accessible to underserved populations worldwide.

Furthermore, ethical considerations and responsible implementation of AI in breast cancer are also crucial. The transparency and explainability of AI algorithms are essential to building trust and ensuring responsible use. AI algorithms must be trained on diverse datasets to avoid bias and ensure equitable access to care for all women. AI should augment human expertise, not replace it. Radiologists and healthcare professionals will continue to play a crucial role in interpreting results, making clinical decisions, and providing compassionate care.

Lastly, as we stand on the brink of a new era in healthcare, the convergence of artificial intelligence and breast cancer detection heralds transformative possibilities. Tools like Mia exemplify how technology can enhance diagnostic accuracy, reduce workloads for healthcare professionals, and, most importantly, save lives through earlier detection. Navigating this landscape requires ensuring these advancements are accessible, equitable, and implemented with the utmost care. The fight against breast cancer is far from over, but with AI as an ally, we are better equipped to face this challenge head-on.