AI model uses histopathological images for accurate diagnosis (representative image)

New Delhi:

According to the World Cancer Report 2022 published by IARC (International Agency for Research on Cancer), breast cancer accounts for 13.6% of all cancer cases (male and female) in India. In women, it goes up to 26 percent of all cancer cases. In the United States, breast cancer accounts for about 30 percent of all new cancer cases in women.

New research shows that artificial intelligence (AI) can help fight this dangerous disease. Early and accurate diagnosis can be critical to treating patients, and a newly developed AI system promises to do just that with near-perfect diagnosis.

A research paper published last month in the journal Cancer, titled “Deep learning-based image classification for breast cancer subtypes and invasiveness assessment from whole-slide image histopathology,” details an AI model that Classifies and identifies different types of breast cancer. The patient, in addition to ruling out malignancy (cancer) in the first place by identifying a benign tumor.

The study — conducted by researchers at Northeastern University, Boston, in collaboration with the Maine Health Institute for Research — developed an AI model that can perform high-resolution histopathological (tissue-level microscopic) whole-slide images of breast tumor tissue. Analyzes.

The AI ​​system, which outperforms previous machine learning (ML) models in the domain by combining predictions from other ML models, uses historical data to classify tumors as malignant (cancerous) or benign (non-cancerous). Worth rating. model during training.

It was trained on publicly available datasets called BreakHis (Breast Cancer Histopathological Database) and BACH (Breast Cancer Histopathology Images). For BACH, microscopic breast tissue images were carefully labeled by clinicians, classifying the images into four classes – normal, benign, in situ carcinoma and invasive carcinoma.

Representative microscopy images showing the four classes in the BACH dataset (image source: Cancer 2024, 16(12), 2222)

And for BreckHis, which consists of 9,109 microscopic images of breast tumor tissue, it was used to further divide benign and malignant tumors into 4 subclasses—each malignant tumor into ductal carcinoma, lobular Carcinoma, mucinous carcinoma and papillary tumorinoma, Fibroadenoma, Phyllodes tumor, and Tubular adenoma.

Representative microscopy images of malignant and benign breast tissues from the BreckHis dataset (image source: Cancer 2024, 16(12), 2222)

Put together, the accuracy of the ensemble ML model is 99.84 percent. Such a performance metric during the research and development phase shows promising promise for real-world application of the technology.

Saeed Amal told Northeastern Global News, “AI can’t miss tumors in a biopsy and won’t stop after 10 or 20 people are diagnosed.” Amal is a professor of bioengineering at Northeastern University and leads the ensemble model project.

In addition to diagnosis, AI systems have also made advances in breast cancer diagnosis and prognosis. For example, AI can now predict breast cancer neoadjuvant chemotherapy (NAC) response using pre-chemotherapy needle biopsy images using hematoxylin and eosin (a common stain in tissue imaging). The AI ​​systems responsible for this have an accuracy of 95.15% and are detailed in a paper titled “Development of multiple AI pipelines for neoadjuvant chemotherapy of breast cancer using H&E-stained tissues. predicts response,” was published in the journal May 2023. of pathology.

In addition, AI has also made significant advances in the identification of lymph node metastases (the spread of cancer cells through the lymph nodes) and the assessment of hormonal status, which is important for breast cancer treatment. These and many other advances made by AI interventions over the years in the fight against breast cancer are described in a review paper published in Diagnostic Pathology in February.



Source link