Breast cancer stands as a significant global health issue, ranking as the most prevalent malignancy among women worldwide. Approximately half a million women die every year due to breast cancer. In 2020, there were 2.3 million women diagnosed with breast cancer and 685,000 deaths globally. Over the last three decades, multiple randomized clinical trials have supported the idea of breast cancer screening being done with mammography. 

Various countries have adopted different approaches to breast cancer screening using mammography. In the United States, women can either initiate the screening process themselves or be referred by their healthcare providers to specialized breast screening centres. Many European screening centres employ a distinctive practice where two radiologists or graders independently assess cases. 

Challenges in current screening methods emphasize the immediate requirement for creative solutions. This necessity is where artificial intelligence (AI) emerges as a significant factor, potentially transforming breast cancer detection and diagnosis. The ability of AI to analyze extensive and intricate datasets surpasses human abilities, potentially recognizing subtle patterns that suggest early-stage cancer and might otherwise go unnoticed by human observers.

Accurate breast cancer detection

Recently, a paper was published by Rentiya Z S, Mandal S, and Inban P on 'Revolutionizing Breast Cancer Detection With AI in Radiology and Radiation Oncology'. The paper explored the current AI landscape in radiology, which features many systems and approaches, each with its unique methodology and output. While a testament to the field's innovation, this diversity poses significant challenges in clinical integration. According to the study, there is a critical need for standardization in how AI systems process data and report findings to ensure consistency, reliability, and interpretability across different platforms and clinical settings. 

Convolutional algorithms for breast cancer screening need to search for soft tissue lesions and calcifications. They must be trained differently than regular convolution algorithms. Initially, at the time of introduction, the convolutional algorithm was trained separately on two different datasets (soft tissue lesions and calcifications), and later on, it was combined. 

Using AI

AI can be used in managing and treating breast cancer through surgical management, radiation oncology and medical oncology. 

The applications of AI in breast surgery offers significant potential, prompting surgeons to remain updated on advancements to enhance their clinical methods, thereby improving patient results. Artificial intelligence tools like chatbots have shown convincing pre-, intra-, and postoperative breast reconstruction results. The integration of AI in these operations could play a significant role in the comprehensive approach to treating breast cancer, focusing on restoring a patient's body image and improving quality of life post-surgeries like mastectomy and lumpectomy. Artificial intelligence's potential in breast cancer screening can significantly reduce error rates and physician workload.

Artificial intelligence can support radiation oncologists by aiding in preparing radiation therapy delivery, leveraging data from various sources like images, treatment algorithms, and dose-volume parameters, enhancing this crucial aspect of breast cancer treatment. As mentioned above, AI algorithms aid in interpreting mammograms, thereby offering improved accuracy in identifying lesions. Furthermore, AI can also help predict treatment responses, assisting clinicians in devising personalized therapy plans based on individual patient characteristics and tumour behaviour.

A study by Dodington et al. utilized AI to analyze nuclear features in breast cancer biopsies to predict response to neoadjuvant chemotherapy. Neoadjuvant chemotherapy is generally administered before surgery to shrink tumours. Predicting a patient's response to this treatment is crucial, as it helps doctors personalize therapy and improve outcomes. 

Future of Mammography and AI

While digital mammography has been advantageous in the realm of breast cancer screening, there are limitations to its use. These include overdiagnosis, overtreatment, and false-positive rates with associated psychological impact and unnecessary costs and biopsies. A potential area of interest for AI development in breast cancer detection that needs further work is mammographic risk assessment metrics.

The radiologist and the AI system may overlook an equal number of cases in a breast cancer screening population. Whether this poses a significant problem hinges on the type of breast cancer detected and missed by both parties. Further assessment is imperative to ascertain the long-term implications. Moreover, addressing the question of responsibility from a legal standpoint necessitates the implementation of quality control protocols for AI algorithms and regular performance audits conducted by radiologists.

Source: Revolutionizing Breast Cancer Detection With Artificial Intelligence (AI) in Radiology and Radiation Oncology: A Systematic Review

Image source: Unsplash