Prostate cancer tissue classification by multiphoton imaging, automated image analysis and machine learning

Prostate carcinoma, a slow-growing and often indolent tumour, is the second most commonly diagnosed cancer among men worldwide. The prognosis is mainly based on the Gleason system through prostate biopsy analysis. However, new treatment and monitoring strategies depend on a more precise diagnosis. Here, we present results by multiphoton imaging for prostate tumour samples from 120 patients that allow to obtain quantitative parameters leading to specific tumour aggressiveness signatures. An automated image analysis was developed to recognise and quantify stromal fibre and neoplastic cell regions in each image. The set of metrics was able to distinguish between non-neoplastic tissue and carcinoma areas by linear discriminant analysis and random forest with accuracy of 89% +/- 3%, but between Gleason groups of only 46% +/- 6%. The reactive stroma analysis improved the accuracy to 65% +/- 5%, clearly demonstrating that stromal parameters should be considered as additional criteria for a more accurate diagnosis.

Automated summary

Prostate carcinoma, the second most common cancer among men worldwide, is often slow-growing and indolent. Prognosis relies heavily on the Gleason system, but new treatment and monitoring strategies require a more precise diagnosis. This study used multiphoton imaging to analyze prostate tumor samples from 120 patients, resulting in quantitative parameters that can indicate specific tumor aggressiveness signatures. The automated image analysis distinguished between non-neoplastic tissue and carcinoma areas with an accuracy of 89% +/- 3%, but only achieved 46% +/- 6% accuracy in differentiating between Gleason groups. However, the inclusion of stromal parameters improved the accuracy to 65% +/- 5%, highlighting the importance of considering these parameters for a more accurate diagnosis.