On the Reliability of CNNs in Clinical Practice: A Computer-Aided Diagnosis System Case Study
Leukocytes classification is essential to assess their number and status since they are the body’s first defence against infection and disease. Automation of the process can reduce the laborious manual process of review and diagnosis by operators and has been the subject of study for at least two de...
Main Authors: | , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2022-03-01
|
Series: | Applied Sciences |
Subjects: | |
Online Access: | https://www.mdpi.com/2076-3417/12/7/3269 |
_version_ | 1797440526702608384 |
---|---|
author | Andrea Loddo Lorenzo Putzu |
author_facet | Andrea Loddo Lorenzo Putzu |
author_sort | Andrea Loddo |
collection | DOAJ |
description | Leukocytes classification is essential to assess their number and status since they are the body’s first defence against infection and disease. Automation of the process can reduce the laborious manual process of review and diagnosis by operators and has been the subject of study for at least two decades. Most computer-aided systems exploit convolutional neural networks for classification purposes without any intermediate step to produce an accurate classification. This work explores the current limitations of deep learning-based methods applied to medical blood smear data. In particular, we consider leukocyte analysis oriented towards leukaemia prediction as a case study. In particular, we aim to demonstrate that a single classification step can undoubtedly lead to incorrect predictions or, worse, to correct predictions obtained with wrong indicators provided by the images. By generating new synthetic leukocyte data, it is possible to demonstrate that the inclusion of a fine-grained method, such as detection or segmentation, before classification is essential to allow the network to understand the adequate information on individual white blood cells correctly. The effectiveness of this study is thoroughly analysed and quantified through a series of experiments on a public data set of blood smears taken under a microscope. Experimental results show that residual networks perform statistically better in this scenario, even though they make correct predictions with incorrect information. |
first_indexed | 2024-03-09T12:09:27Z |
format | Article |
id | doaj.art-3d0f91d81e824ee581a5e7b6204dc39d |
institution | Directory Open Access Journal |
issn | 2076-3417 |
language | English |
last_indexed | 2024-03-09T12:09:27Z |
publishDate | 2022-03-01 |
publisher | MDPI AG |
record_format | Article |
series | Applied Sciences |
spelling | doaj.art-3d0f91d81e824ee581a5e7b6204dc39d2023-11-30T22:53:25ZengMDPI AGApplied Sciences2076-34172022-03-01127326910.3390/app12073269On the Reliability of CNNs in Clinical Practice: A Computer-Aided Diagnosis System Case StudyAndrea Loddo0Lorenzo Putzu1Department of Mathematics and Computer Science, University of Cagliari, Via Ospedale 72, 09124 Cagliari, ItalyDepartment of Electrical and Electronic Engineering, University of Cagliari, Piazza d’Armi, 09123 Cagliari, ItalyLeukocytes classification is essential to assess their number and status since they are the body’s first defence against infection and disease. Automation of the process can reduce the laborious manual process of review and diagnosis by operators and has been the subject of study for at least two decades. Most computer-aided systems exploit convolutional neural networks for classification purposes without any intermediate step to produce an accurate classification. This work explores the current limitations of deep learning-based methods applied to medical blood smear data. In particular, we consider leukocyte analysis oriented towards leukaemia prediction as a case study. In particular, we aim to demonstrate that a single classification step can undoubtedly lead to incorrect predictions or, worse, to correct predictions obtained with wrong indicators provided by the images. By generating new synthetic leukocyte data, it is possible to demonstrate that the inclusion of a fine-grained method, such as detection or segmentation, before classification is essential to allow the network to understand the adequate information on individual white blood cells correctly. The effectiveness of this study is thoroughly analysed and quantified through a series of experiments on a public data set of blood smears taken under a microscope. Experimental results show that residual networks perform statistically better in this scenario, even though they make correct predictions with incorrect information.https://www.mdpi.com/2076-3417/12/7/3269convolutional neural networkstransfer learningfine-tuningdirect classificationblood smear imagesleukaemia diagnosis |
spellingShingle | Andrea Loddo Lorenzo Putzu On the Reliability of CNNs in Clinical Practice: A Computer-Aided Diagnosis System Case Study Applied Sciences convolutional neural networks transfer learning fine-tuning direct classification blood smear images leukaemia diagnosis |
title | On the Reliability of CNNs in Clinical Practice: A Computer-Aided Diagnosis System Case Study |
title_full | On the Reliability of CNNs in Clinical Practice: A Computer-Aided Diagnosis System Case Study |
title_fullStr | On the Reliability of CNNs in Clinical Practice: A Computer-Aided Diagnosis System Case Study |
title_full_unstemmed | On the Reliability of CNNs in Clinical Practice: A Computer-Aided Diagnosis System Case Study |
title_short | On the Reliability of CNNs in Clinical Practice: A Computer-Aided Diagnosis System Case Study |
title_sort | on the reliability of cnns in clinical practice a computer aided diagnosis system case study |
topic | convolutional neural networks transfer learning fine-tuning direct classification blood smear images leukaemia diagnosis |
url | https://www.mdpi.com/2076-3417/12/7/3269 |
work_keys_str_mv | AT andrealoddo onthereliabilityofcnnsinclinicalpracticeacomputeraideddiagnosissystemcasestudy AT lorenzoputzu onthereliabilityofcnnsinclinicalpracticeacomputeraideddiagnosissystemcasestudy |