# Reg R-CNN: Lesion Detection and Grading under Noisy Labels

## TL;DR

This paper introduces Reg R-CNN, a novel object detection framework for medical images that replaces classification with regression to better handle continuous labels and noisy data, improving lesion detection and grading accuracy.

## Contribution

Reg R-CNN innovatively replaces the classification stage with a regression model in object detection, effectively managing continuous labels and label noise in medical imaging.

## Key findings

- Outperforms traditional classifiers on lesion grading tasks
- Effective in handling noisy and ambiguous labels
- Demonstrates superiority on a public medical dataset

## Abstract

For the task of concurrently detecting and categorizing objects, the medical imaging community commonly adopts methods developed on natural images. Current state-of-the-art object detectors are comprised of two stages: the first stage generates region proposals, the second stage subsequently categorizes them. Unlike in natural images, however, for anatomical structures of interest such as tumors, the appearance in the image (e.g., scale or intensity) links to a malignancy grade that lies on a continuous ordinal scale. While classification models discard this ordinal relation between grades by discretizing the continuous scale to an unordered bag of categories, regression models are trained with distance metrics, which preserve the relation. This advantage becomes all the more important in the setting of label confusions on ambiguous data sets, which is the usual case with medical images. To this end, we propose Reg R-CNN, which replaces the second-stage classification model of a current object detector with a regression model. We show the superiority of our approach on a public data set with 1026 patients and a series of toy experiments. Code will be available at github.com/MIC-DKFZ/RegRCNN.