Incorporating clinical metadata with digital image features for automated identification of cutaneous melanoma
Liu, Zhao; Sun, Jiuai; Smith, Melvyn; Smith, Lyndon
Jiuai Sun firstname.lastname@example.org
Melvyn Smith Melvyn.Smith@uwe.ac.uk
Research Centre Director Vision Lab/Prof
Lyndon Smith Lyndon.Smith@uwe.ac.uk
Professor in Computer Simulation and Machine
Background Computer-assisted diagnosis (CAD) of malignant melanoma (MM) has been advocated to help clinicians in achieving a more objective and reliable assessment. However,
conventional CAD systems only examine the features extracted from digital photographs of lesions. Failure to incorporate patient personal information constrains their applicability in clinical settings.
Objectives To develop a new CAD system to improve the performance of automatic diagnosis of melanoma which, for the first time, incorporates lesion digital features with
important patient metadata into learning process.
Methods Thirty-two features are extracted from digital photographs to characterise skin lesions. Patient personal information like age, gender, lesion site and their combinations are quantified as metadata. The integration of digital features and metadata is realised through an
extended Laplacian Eigenmap, a dimensionality reduction method grouping lesions with similar digital features and metadata into the same classes.
Results The diagnosis is 82.14% sensitivity and 86.07% specificity, when only multidimensional digital features are used; while the results are significantly improved to
95.25% sensitivity and 90.96% specificity respectively, after metadata are incorporated appropriately. The proposed system achieves a level of sensitivity comparable to experienced dermatologists aided by conventional dermoscopes. This demonstrates the potential of our
method for assisting clinicians in diagnosing melanoma, and the benefit it could provide to patients and hospitals by greatly reducing unnecessary excision of benign naevi.
Conclusions This paper proposes an enhanced CAD system by incorporating clinical metadata in the learning process for automatic classification of melanoma. Results demonstrate that the additional metadata and the mechanism to incorporate them are useful for improving computer-assisted diagnosis of melanoma.
Liu, Z., Sun, J., Smith, M., & Smith, L. (2013). Incorporating clinical metadata with digital image features for automated identification of cutaneous melanoma. British Journal of Dermatology,
|Journal Article Type||Article|
|Publication Date||Jan 1, 2013|
|Journal||British Journal of Dermatology|
|Peer Reviewed||Peer Reviewed|
|Keywords||computer-assisted diagnosis, clinical metadata, digital image analysis, malignant melanoma|
This file is under embargo due to copyright reasons.
You might also like
Weed classification in grasslands using convolutional neural networks
Broad-leaf weed detection in pasture
Multispectral contactless 3D handprint acquisition for identification
Presentation / Conference
2D and 3D face analysis for ticketless rail travel