Health

Artificial intelligence and diagnosis of prostate cancer

Read the article

AIRAMatrix CEO Chaith Kondragunta talks about AI-based prostate cancer diagnosis

Prostate cancer is the second most common cancer among men and the fifth most common cause of cancer death. The illness ranges from lazy and slow-growing (only careful follow-up required) to aggressive and deadly illness (requiring immediate therapeutic intervention). Therefore, it guarantees a coordinated treatment plan. In addition, underdiagnosis can lead to increased mortality from illness, and overdiagnosis can lead to unnecessary treatment, resulting in morbidity due to side effects. Therefore, accurate disease stratification and coordinated treatment decisions are just as important as early disease detection and diagnosis for effective treatment of prostate cancer.

Unfortunately, early detection of prostate cancer is possible, but early identification is not possible. This creates uncertainty about disease progression and treatment decisions. Therefore, in addition to sensitive and reasonably specific screening methods for early disease detection and diagnosis, techniques for improving disease stratification and prognosis that help establish appropriate treatment regimens are available. , Is the basis of prostate cancer treatment pathways. AIRA Matrix aims to have a positive impact on prostate cancer outcomes by applying AI-based solutions to the diagnosis, prognosis, and prediction of the entire care path, from screening to postoperative follow-up.

The Gleason score is an important histological parameter that stratifies cancer into groups of different grades and assists in therapeutic decisions. This is currently determined by a professional pathologist through a microscopic examination of the biopsy tissue. Between observers in Gleason classification, especially at key clinical decision points such as Gleason Grade Group 1 (which may require only active monitoring) and Gleason Grade Group 2 (which may require therapeutic intervention). Variability is widely recognized and is compounded by a lack of expertise in the required disciplines, often inconsistent disease stratification to achieve optimal grading accuracy. be connected.

These types of problems are well suited for applying deep learning techniques. We have developed a new deep learning-based solution for pixel-level semantic segmentation. It closely follows the approach applied by experienced pathologists and offers better performance than current approaches. This approach also reduced the variability between observers in reporting histopathological images and improved the accuracy, accuracy, and speed of Gleason’s scoring. We believe that these improvements will enable a better prognosis for prostate cancer and help guide the selection of appropriate treatment regimens.

Beyond the Gleason grading, we are currently working on solutions that predict the course, spread, and aggression of cancer. Our goal is to improve prostate cancer screening, prognosis, treatment planning, and follow-up with better results for patients and caregivers.

Prostate cancer treatment pathways span disease stratification using biopsy and other early disease detection and diagnosis studies, biochemical, radiological and histopathological parameters, and careful follow-up Grade and stage the disease to determine treatment options such as active monitoring, definitive surgical intervention, and chemotherapy. Radiation therapy, and careful follow-up of treated patients, were grouped into these different treatment regimen groups.

Our products and solutions aim to work at every stage of the prostate cancer care pathway to be more effective and outcome-focused. In the screening process, triage solutions help expedite screening of prostate needle core biopsies to detect and diagnose malignancies. The algorithm works in two ways: screening and triaging the biopsy section to point out areas of importance to the pathologist, or as a second read, identifying tumor areas that the pathologist may have overlooked. To do. A dramatic increase in the number of CNBs reviewed on a case-by-case basis over the last decade can increase the workload of pathologists, delay cancer diagnosis, and lead to diagnostic errors due to fatigue. I have. Our solution aims to reduce this burden on pathologists.

Our solution also helps in the risk stratification step by annotating each core biopsy by gland and automatically calculating Gleason grades and scores. The goal is to reduce subjectivity and variability between observers.

Finally, we aim to leverage information from the gold standard of histopathology to enhance other non-invasive modalities such as radiology and molecular testing. Example: Multiparametric magnetic resonance imaging (mpMRI) is an important diagnostic modality. It is used for primary diagnosis, risk stratification, and treatment of prostate cancer patients. However, mpMRI significantly underestimates size, extent, and tumor volume compared to histopathology. We are working on 3D reconstruction of tumor specimens by simultaneous registration of mpMRI and radical prostatectomy images. We hope to improve staging, topical treatment planning, and follow-up with these less invasive procedures.

We believe that these solutions will enable appropriate and timely treatment options and improve the outcome of the entire prostate treatment pathway.

Artificial intelligence and diagnosis of prostate cancer

Source link Artificial intelligence and diagnosis of prostate cancer

Related Articles

Back to top button