DNA damage analysis can provide valuable information in several areas ranging from the diagnosis/treatment of a disease to the monitoring of the effects of genetic and environmental influences. The evaluation of the damage is determined by comet scoring, which can be performed by a skilled operator with a manual procedure. However, this approach becomes very time-consuming and the operator dependency results in the subjectivity of the damage quantification and thus in a high inter/intra-operator variability.

Methods:
In this paper, we aim to overcome this issue by introducing a Deep Learning methodology based on Faster R-CNN to completely automatize the overall approach while discovering unseen discriminative patterns in comets.

Results:
The experimental results performed on two real use-case datasets reveal the higher performance (up to mean absolute precision of 0.74) of the proposed methodology against other state-of-the-art approaches. Additionally, the validation procedure performed by expert biologists highlights how the proposed approach is able to unveil true comets, often unseen from the human eye and standard computer vision methodology.

Conclusions:
This work contributes to the biomedical informatics field by the introduction of a novel approach based on established object detection Deep Learning technique for evaluating the DNA damage. The main contribution is the application of Faster R-CNN for the detection and quantification of DNA damage in comet assay images, by fully automatizing the detection/classification DNA damage task. The experimental results extracted in two real use-case datasets demonstrated (i) the higher robustness of the proposed methodology against other state-of-the-art Deep Learning competitors, (ii) the speeding up of the comet analysis procedure and (iii) the minimization of the intra/inter-operator variability.

Introduction
The single cell gel electrophoresis (SCGE) assay, also known as the Comet Assay, is a technique used in medical and biological fields for evaluating DNA damage in individual cells[1]. DNA damage analysis can provide valuable information in several areas ranging from the diagnosis/treatment of a disease to the monitoring of the effects of genetic and environmental influences or hazardous chemical exposure[2], [3], [4]. From microscope images, it is possible to observe that, when damaged, DNA resembles a comet, with two main regions: a circular head composed of intact DNA and a tail of damaged one. Usually, the evaluation of the damage is determined by comet scoring[5], which can be performed by a skilled operator with a manual procedure. Even if the expert is capable of assessing DNA damage based on comets’ morphological features by visual inspection, this approach becomes very time-consuming in case of a huge amount of images[6]. Moreover, the operator dependency results in the subjectivity of the damage quantification and thus in a high inter/intra-operator variability[7]. In order to overcome these issues, computer-based image analysis tools were developed to support comet assay procedure and they can be mainly classified as semi-automated and fully-automated software. Semi-automated tools speed up the analysis procedure compared to visual inspection, but typically they require the interaction of an expert across different stages to manually set several parameters[8]. Instead, fully automated software employ Computer Vision and/or Machine Learning (ML) techniques, avoiding user dependency and thus providing standardized and reproducible measurements. However, for both strategies the detection/classification tasks still present several challenges mainly due to (i) the quality of the images (ii) the overlapping of different comets and (iii) the difficulty to classify multi-level of damage. As a consequence of these limitations, these approaches employ hand-crafted features and the overall method is not at all fully automated (i.e.,the experimental procedure and the features extraction stage is driven by the supervision of an expert operator).

Read all the article >