Back

A simple computer vision algorithm as a clinical aid for the pathologist

Lazaro-Guevara, J. M.; Garrido, K. M.; Flores-Robles, B. J.; Murga, A. E.

2019-11-25 bioinformatics
10.1101/853325 bioRxiv
Show abstract

Histological analysis for cancer detection or stratification is performed by observing and examining a small portion of a biopsied tissue under a microscope. Nevertheless, to assign clinical meaning to the findings, the analysis and interpretation of an experienced Pathologist is always necessary. Using high-resolution images, these experts visually examine the sample looking for specific characteristics on the cell shapes and tissue distributions, so they could decide whether tissue regions are cancerous, and establish the malignancy level of it. However, with the increasing demand for work for those pathologists and the importance of accuracy on diagnostics, multiple attempts to simplify their work have been performed. Current Imaging technologies allow novel horizons in the automatized selection of some of the characteristics that indicate malignancy in a biopsy. In this work, we propose a simple computer vision algorithm that can be implemented as a screening method for focusing in histological areas with higher risk of malignancy saving time to the pathologist and helping to perform a more standardized work, an easy observation with the potential to become in an aid to daily clinical work.

Matching journals

The top 6 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.