A new artificial intelligence software has now been authorized by the US-FDA, which will help the pathologists in the detection of sites suspicious for cancer in addition to the review of digitally-scanned slide images obtained from prostate biopsies, a press release on 21st September 2021 revealed. This is the first of its kind software that recognizes the area of interest on the biopsy image with the maximum possibility of cancer invasion if this area has been missed earlier.

This software is compatible for usage with digitized slide images that can be later envisaged via a slide image viewer. This software authorization is based on the clinical study comprising of 16 pathologists who assessed 527 digitized slide images of prostate biopsies (356 benign;171 cancer). As compared to the assessment without the software used (unassisted read), the use of this software significantly improved the revelation of cancer on individual slide images by 7.3% on average. There was no significant impact on interpretation of benign slide images.

There was a risk of false positive and false negative outcomes, that is alleviated by the device’s use as an adjunct and by assessment by a skilled pathologist based on the history of the patients along with other clinical data, and who might carry out additional lab studies on the samples before a final verdict. This software analysis is in accordance with a regulatory pathway for moderate to low risk devices of the latest category called the De Novo process.

The FDA is also establishing special controls for these devices, including prerequisites concerning labelling and performance testing along with authorization. After the fulfilment of these controls (special plus general), the safety and effectiveness of these devices can be determined. This step will further enable a new regulatory categorization where the devices of the same type with the same proposed use may go through a 510(k) premarket process by the FDA by acquiring marketing authorization with reference to a former device of the same kind.