Lagatar24 Desk
New Delhi: Breast cancer accounts for 13.6% of all cancer cases in India, according to the 2022 World Cancer Report by IARC (International Agency for Research on Cancer). Among women, this figure rises to 26%. In the United States, breast cancer represents about 30% of all new cancer cases among women. Recent research indicates that Artificial Intelligence (AI) could play a crucial role in combating this pervasive disease by enabling early and precise diagnosis.
A recent study titled “Ensemble Deep Learning-Based Image Classification for Breast Cancer Subtype and Invasiveness Diagnosis from Whole Slide Image Histopathology,” published in the Cancers journal, highlights an AI model capable of accurately classifying and identifying various types of breast cancer. This AI system also distinguishes between malignant and benign tumors, offering a promising tool for early detection and treatment.
Researchers from Northeastern University, Boston, and the Maine Health Institute for Research developed this AI model, which analyzes high-resolution histopathological images of breast tumor tissue. By combining predictions from multiple machine learning (ML) models, this ensemble AI system surpasses previous ML models in accuracy. It was trained on publicly available datasets, including the Breast Cancer Histopathological Database (BreakHis) and the Breast Cancer Histopathology Images (BACH).
The BreakHis dataset consists of 9,109 microscopic images of breast tumor tissue, which were used to categorize tumors into four malignant subtypes (ductal carcinoma, lobular carcinoma, mucinous carcinoma, and papillary carcinoma) and four benign subtypes (adenosis, fibroadenoma, phyllodes tumor, and tubular adenoma). The BACH dataset features images meticulously labeled by medical experts into four categories: normal, benign, in situ carcinoma, and invasive carcinoma.
The ensemble AI model demonstrated an impressive accuracy of 99.84% during research and development, showing significant potential for real-world applications. “The AI can’t miss a tumor in the biopsy and won’t be exhausted after diagnosing 10 or 20 people,” said Saeed Amal, a bioengineering professor at Northeastern University and the project lead, to Northeastern Global News.
In addition to diagnosis, AI systems have made advancements in prognosis and treatment predictions for breast cancer. AI can now predict the response to neoadjuvant chemotherapy (NAC) using Hematoxylin and eosin-stained tissue images from pre-chemotherapy needle biopsies. This AI system, described in the paper “Development of multiple AI pipelines that predict neoadjuvant chemotherapy response of breast cancer using H&E-stained tissues,” published in the Journal of Pathology in May 2023, achieves an accuracy of 95.15%.
Moreover, AI has shown significant progress in identifying lymph node metastasis (the spread of cancer cells through lymph nodes) and evaluating hormonal status, which are critical for breast cancer treatment. These advancements, among others, were reviewed in a paper published in Diagnostic Pathology in February.
