The seven key challenges for the future of computer-aided diagnosis in medicine
Introduction
The motivation for the present study became apparent after a recent systematic survey that the authors did on the subject of computer-aided diagnosis (CAD) in medicine regarding its past and current developments [1]. The first study indicated that there are some challenges still facing CAD development that seriously inhibit its progress. These challenges have to be understood well in order to bring CAD to the next level. By addressing these challenges, it is possible to eliminate existing barriers in CAD application and development.
Fig. 1 presents a brief history of the key advances in this area [2,57,[102], [103], [104], [105], [106], [107],109,110,112,113]. At the beginning of the early studies in CAD, researchers were hoping to develop entirely automatic computer-aided diagnostic systems. However, thanks to the better understanding reached in the early 1970s [2] of what computers can do or cannot do, this attitude has sifted. Currently, CAD is considered more as a diagnostic aid tool for physicians, as opposed to a completely automatic system for making medical diagnoses [3].
As CAD spans in the interface of medicine and computer science (see also Fig. 2), the challenges that CAD faces today are also related to these two constituent fields. Therefore, if progress in one of these fields occurs in relation to CAD, it is very likely that the other constituent field will benefit as well.
Section snippets
Methods
The initial literature search protocol followed was the well-accepted PRISMA approach [4]. The traditional PRISMA approach comes with some variants depending on the goal of the study (see also www.prisma-statement.org). However, existing PRISMA approaches are static, as they assume that the researchers already have a comprehensive understanding of all the issues involved (and hence the relevant keywords).
The database under consideration for the initial search was PubMed. The keywords / phrases,
Results and Discussion
The literature searches for CAD related publications have uncovered a number of challenges that modern CAD systems face. A typical CAD system, like most standard data mining systems, usually follows a four-step approach [6]. These steps are ordered as follows: data acquisition, data pre-processing, main processing, and visualization / presentation of the results. Moreover, the main processing step of CAD systems often involves three sub-steps. These sub-steps are as follows: data segmentation,
Conclusions
CAD systems play a significant role assisting physicians making objective and effective diagnostic decisions. This paper introduced a new literature review protocol, called the dynamic PRISMA approach. This approach is an enhancement of the traditional PRISMA approach used in such cases. Next it identified seven key challenges that CAD systems face today. This paper also described recent research efforts towards these seven challenges as well as it identified areas where future work has the
Contributors
The two authors conceived and designed the study. They drafted the manuscript. Both authors critically revised the work for important intellectual content, and approved the final version of the manuscript for publication.
Funding
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
The originality of the submitted manuscript
This is an original paper and has not been submitted or presented anywhere else.
We also declare that there is no case of plagiarism in this manuscript.
Conflict of Interest Statement
There is no case of conflict of interest with any of the two co-authors of this manuscript.
Acknowledgments
The authors of this paper are very appreciative to the Editor-in-Chief Professor Heimar de Fátima Marin and the two anonymous reviewers the comments of which were valuable in clarifying and enhancing the original version of this paper.
References (114)
Computer-aided diagnosis in medical imaging: historical review, current status and future potential
Comput. Med. Imaging Graph.
(2007)- et al.
Efficient multi-scale 3D CNN with fully connected CRF for accurate brain lesion segmentation
Med. Image Anal.
(2017) - et al.
Cardiac sarcoidosis classification with deep convolutional neural network-based features using polar maps
Comput. Biol. Med.
(2019) - et al.
Three-dimensional Gabor feature extraction for hyperspectral imagery classification using a memetic framework
Inf. Sci. (Ny)
(2015) - et al.
Automatic brain tissue segmentation in MR images using random forests and conditional random fields
J. Neurosci. Methods
(2016) - et al.
Evaluating imaging and computer-aided detection and diagnosis devices at the FDA
Acad. Radiol.
(2012) - et al.
Standardized evaluation of algorithms for computer-aided diagnosis of dementia based on structural MRI: the CADDementia challenge
NeuroImage
(2015) - et al.
CAD: how it works, how to use it, performance
Eur. J. Radiol.
(2013) Fuzzy sets
Inf. Control.
(1965)Intuitionistic fuzzy sets
Fuzzy Sets Syst.
(1986)
Development and evaluation of five fuzzy multiattribute decision-making methods
Int. J. Approx. Reason.
Fuzzy logic in computer-aided breast cancer diagnosis: analysis of lobulation
Artif. Intell. Med.
Intuitionistic fuzzy C-regression by using least squares support vector regression
Expert Syst. Appl.
A hybrid metaheuristic and kernel intuitionistic fuzzy c-means algorithm for cluster analysis
Appl. Soft Comput.
Intuitionistic fuzzy information–applications to pattern recognition
Pattern Recognit. Lett.
IBM Watson: how cognitive computing can be applied to big data challenges in life sciences research
Clin. Ther.
Evaluation of computer-aided detection devices
Acad. Radiol.
A systematic survey of computer-aided diagnosis in medicine: past and present developments
Expert Syst. Appl.
Reducibility among combinatorial problems
Complexity of Computer Computations
Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement
Ann. Intern. Med.
What I learned from predatory publishers
Biochem. Med. (Zagreb)
Toward reuse of clinical data for research and quality improvement: the end of the beginning?
Ann. Intern. Med.
A national survey assessing the number of records allowed open in electronic health records at hospitals and ambulatory sites
J. Am. Med. Inform. Assoc.
Hospital network hacked
4.5 Million Records Stolen
Next-generation phenotyping of electronic health records
J. Am. Med. Inform. Assoc.
Impact of electronic health record systems on information integrity: quality and safety implications
Perspect. Health Inf. Manag.
Methods and dimensions of electronic health record data quality assessment: enabling reuse for clinical research
J. Am. Med. Inform. Assoc.
Training clinicians in how to use patient-reported outcome measures in routine clinical practice
Qual. Life Res.
A new EHR training curriculum and assessment for pediatric residents
Appl. Clin. Inform.
How to Investigate the Use of Medicines by Consumers
A theory of the learnable
Commun. ACM
Identifying and avoiding bias in research
Plast. Reconstr. Surg.
Image Processing, Analysis, and Machine Vision
Computer-aided diagnosis: how to move from the laboratory to the clinic
Radiology
Automatic liver segmentation using an adversarial image-to-image network
International Conference on Medical Image Computing and Computer-Assisted Intervention
Content-based medical image retrieval: a survey of applications to multidimensional and multimodality data
J. Digit. Imaging
An ensemble of fine-tuned convolutional neural networks for medical image classification
IEEE J. Biomed. Health Inform.
Rich feature hierarchies for accurate object detection and semantic segmentation
Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition
Fully convolutional networks for semantic segmentation
Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition
Recurrent instance segmentation
European Conference on Computer Vision, October Springer, Cham
Deep learning
Nature
Brain tumor detection and segmentation in a CRF (conditional random fields) framework with pixel-pairwise affinity and superpixel-level features
Int. J. Comput. Assist. Radiol. Surg.
October. Automatic liver and lesion segmentation in CT using cascaded fully convolutional neural networks and 3D conditional random fields
International Conference on Medical Image Computing and Computer-Assisted Intervention
Deep convolutional neural networks for computer-aided detection: CNN architectures, dataset characteristics and transfer learning
IEEE Trans. Med. Imaging
Mask r-cnn
Computer Vision (ICCV), 2017 IEEE International Conference, October, IEEE
Chest pathology identification using deep feature selection with non-medical training
Comput. Methods Biomech. Biomed. Eng. Imaging Vis.
Feature extraction using convolutional neural network for classifying breast density in mammographic images
Medical Imaging 2017: Computer-Aided Diagnosis (Vol. 10134, p. 101342M)
Automatic segmentation of MR brain images with a convolutional neural network
IEEE Trans. Med. Imaging
Cited by (46)
Automated image label extraction from radiology reports — A review
2024, Artificial Intelligence in MedicineDetection of dental restorations using no-code artificial intelligence
2023, Journal of DentistryBig data in healthcare: Conceptual network structure, key challenges and opportunities
2023, Digital Communications and NetworksUsing feature maps to unpack the CNN ‘Black box’ theory with two medical datasets of different modality
2023, Intelligent Systems with ApplicationsDisease Localization and Severity Assessment in Chest X-Ray Images using Multi-Stage Superpixels Classification
2022, Computer Methods and Programs in BiomedicineCitation Excerpt :While subdivided the lung fields, the number of non-infected superpixels is significantly larger than that of infected superpixels, which substantially affects the performance. Moreover, to train a generalized model, a balanced dataset is required [65]. The results of this study demonstrate that infected region in CXR images can be efficiently identified using multistage superpixel classification approach.
Artificial neural network-based approaches for computer-aided disease diagnosis and treatment
2022, Cognitive and Soft Computing Techniques for the Analysis of Healthcare Data