Discriminative Scale Learning (DiScrn): Applications to Prostate Cancer Detection from MRI and Needle Biopsies

Haibo Wang,Anant Madabhushi,Satish Viswanath

doi:10.1038/s41598-017-12569-z

Haibo Wang, Anant Madabhushi + Show 1 more

Open Access

https://doi.org/10.1038/s41598-017-12569-z

Copy DOI

Abstract

There has been recent substantial interest in extracting sub-visual features from medical images for improved disease characterization compared to what might be achievable via visual inspection alone. Features such as Haralick and Gabor can provide a multi-scale representation of the original image by extracting measurements across differently sized neighborhoods. While these multi-scale features are effective, on large-scale digital pathological images, the process of extracting these features is computationally expensive. Moreover for different problems, different scales and neighborhood sizes may be more or less important and thus a large number of features extracted might end up being redundant. In this paper, we present a Discriminative Scale learning (DiScrn) approach that attempts to automatically identify the distinctive scales at which features are able to best separate cancerous from non-cancerous regions on both radiologic and digital pathology tissue images. To evaluate the efficacy of our approach, our approach was employed to detect presence and extent of prostate cancer on a total of 60 MRI and digitized histopathology images. Compared to a multi-scale feature analysis approach invoking features across all scales, DiScrn achieved 66% computational efficiency while also achieving comparable or even better classifier performance.

Highlights

Selection tend to rely heavily on scale sampling such as dense sampling[12] or ad hoc sampling[14]
We present a new Discriminative Scale learning (DiScrn) based approach to tackle the problem of selecting discriminating scales for multi-scale feature extraction from medical images
We begin by noting that the objective of the experiments in this study was not that DiScrn yields the best possible prostate cancer detection classifier on MRI and histopathology

Summary

Introduction

Selection tend to rely heavily on scale sampling such as dense sampling[12] or ad hoc sampling[14]. By deformably registering the in vivo imaging with the ex vivo pathology we are able to spatially map disease extent onto the in vivo imaging This “ground truth” mapping for disease extent allows for training and evaluating the discriminative scale based learning approach for cancer diagnosis. To robustly evaluate our approach we used data from two different institutions, using data from one site to train and data from the other site to validate the DiScrn approach This is, to the best of our knowledge, the first instance of an attempt to use data from different sites to train and validate a computer aided diagnosis classifier for prostate cancer from multi-parametric MRI.

Objectives

Methods

Results

Conclusion