The Impact of the Observation of Predictive Features on the Diagnosis of Pigmented Skin Lesions and the therapeutic decision

Abstract : In this paper, we study the relationship between diagnosis and therapeutic decision on the one hand and the observations of the presence of ABCD features and some additional dermoscopic features of pigmented skin tumors on the other hand. The image database was composed of 227 images of pigmented skin lesions. Five senior dermatologists were asked for their expertise about these images. They gave their opinion about the presence of ABCD and dermoscopic features, their diagnosis and their therapeutic decision. The performances of dermatologists were evaluated in terms of their ability to diagnose melanoma by building statistical decision models from their observations of predictive features. Models allowed observing to what extent dermatologists ground their diagnosis on the malignancy features they detected. It appeared that a high variability of behavior among dermatologists is observed, concerning both the detection of features and the role of features for the elaboration of diagnosis.
Document type :
Conference papers
Complete list of metadatas

Cited literature [8 references]  Display  Hide  Download

https://hal-supelec.archives-ouvertes.fr/hal-00756094
Contributor : Alexandra Siebert <>
Submitted on : Thursday, November 22, 2012 - 2:55:30 PM
Last modification on : Wednesday, May 15, 2019 - 3:48:14 AM
Long-term archiving on : Saturday, February 23, 2013 - 3:45:11 AM

File

final_rits_skinan_v1.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-00756094, version 1

Citation

Y. Wazaefi, Arthur Tenenhaus, Alex Nkengne, Jean-Francois Horn, Alain Giron, et al.. The Impact of the Observation of Predictive Features on the Diagnosis of Pigmented Skin Lesions and the therapeutic decision. DMIN'12, Jul 2012, Nevada, United States. CD-ROM Proceedings (4 p.). ⟨hal-00756094⟩

Share

Metrics

Record views

1064

Files downloads

214