Deep Neural Frameworks Improve the Accuracy of General Practitioners in the Classification of Pigmented Skin Lesions
This study evaluated whether deep learning frameworks trained in large datasets can help non-dermatologist physicians improve their accuracy in categorizing the seven most common pigmented skin lesions. Open-source skin images were downloaded from the International Skin Imaging Collaboration (ISIC)...
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2020-11-01
|
| Series: | Diagnostics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2075-4418/10/11/969 |
| _version_ | 1797547545198592000 |
|---|---|
| author | Maximiliano Lucius Jorge De All José Antonio De All Martín Belvisi Luciana Radizza Marisa Lanfranconi Victoria Lorenzatti Carlos M. Galmarini |
| author_facet | Maximiliano Lucius Jorge De All José Antonio De All Martín Belvisi Luciana Radizza Marisa Lanfranconi Victoria Lorenzatti Carlos M. Galmarini |
| author_sort | Maximiliano Lucius |
| collection | DOAJ |
| description | This study evaluated whether deep learning frameworks trained in large datasets can help non-dermatologist physicians improve their accuracy in categorizing the seven most common pigmented skin lesions. Open-source skin images were downloaded from the International Skin Imaging Collaboration (ISIC) archive. Different deep neural networks (DNNs) (<i>n</i> = 8) were trained based on a random dataset constituted of 8015 images. A test set of 2003 images was used to assess the classifiers’ performance at low (300 × 224 RGB) and high (600 × 450 RGB) image resolution and aggregated data (age, sex and lesion localization). We also organized two different contests to compare the DNN performance to that of general practitioners by means of unassisted image observation. Both at low and high image resolution, the DNN framework differentiated dermatological images with appreciable performance. In all cases, the accuracy was improved when adding clinical data to the framework. Finally, the least accurate DNN outperformed general practitioners. The physician’s accuracy was statistically improved when allowed to use the output of this algorithmic framework as guidance. DNNs are proven to be high performers as skin lesion classifiers and can improve general practitioner diagnosis accuracy in a routine clinical scenario. |
| first_indexed | 2024-03-10T14:45:34Z |
| format | Article |
| id | doaj.art-29988baf95204117885943a84b9b11cb |
| institution | Directory Open Access Journal |
| issn | 2075-4418 |
| language | English |
| last_indexed | 2024-03-10T14:45:34Z |
| publishDate | 2020-11-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Diagnostics |
| spelling | doaj.art-29988baf95204117885943a84b9b11cb2023-11-20T21:24:15ZengMDPI AGDiagnostics2075-44182020-11-01101196910.3390/diagnostics10110969Deep Neural Frameworks Improve the Accuracy of General Practitioners in the Classification of Pigmented Skin LesionsMaximiliano Lucius0Jorge De All1José Antonio De All2Martín Belvisi3Luciana Radizza4Marisa Lanfranconi5Victoria Lorenzatti6Carlos M. Galmarini7Topazium Artificial Intelligence, Paseo de la Castellana 40 Pl 8, 28046 Madrid, SpainSanatorio Otamendi, C1115AAB Buenos Aires, ArgentinaSanatorio Otamendi, C1115AAB Buenos Aires, ArgentinaTopazium Artificial Intelligence, Paseo de la Castellana 40 Pl 8, 28046 Madrid, SpainInstituto de Obra Social de las Fuerzas Armadas, C1115AAB Buenos Aires, ArgentinaSanatorio Otamendi, C1115AAB Buenos Aires, ArgentinaSanatorio Otamendi, C1115AAB Buenos Aires, ArgentinaTopazium Artificial Intelligence, Paseo de la Castellana 40 Pl 8, 28046 Madrid, SpainThis study evaluated whether deep learning frameworks trained in large datasets can help non-dermatologist physicians improve their accuracy in categorizing the seven most common pigmented skin lesions. Open-source skin images were downloaded from the International Skin Imaging Collaboration (ISIC) archive. Different deep neural networks (DNNs) (<i>n</i> = 8) were trained based on a random dataset constituted of 8015 images. A test set of 2003 images was used to assess the classifiers’ performance at low (300 × 224 RGB) and high (600 × 450 RGB) image resolution and aggregated data (age, sex and lesion localization). We also organized two different contests to compare the DNN performance to that of general practitioners by means of unassisted image observation. Both at low and high image resolution, the DNN framework differentiated dermatological images with appreciable performance. In all cases, the accuracy was improved when adding clinical data to the framework. Finally, the least accurate DNN outperformed general practitioners. The physician’s accuracy was statistically improved when allowed to use the output of this algorithmic framework as guidance. DNNs are proven to be high performers as skin lesion classifiers and can improve general practitioner diagnosis accuracy in a routine clinical scenario.https://www.mdpi.com/2075-4418/10/11/969artificial intelligencedermatologydeep learningskin diseasesmelanoma |
| spellingShingle | Maximiliano Lucius Jorge De All José Antonio De All Martín Belvisi Luciana Radizza Marisa Lanfranconi Victoria Lorenzatti Carlos M. Galmarini Deep Neural Frameworks Improve the Accuracy of General Practitioners in the Classification of Pigmented Skin Lesions Diagnostics artificial intelligence dermatology deep learning skin diseases melanoma |
| title | Deep Neural Frameworks Improve the Accuracy of General Practitioners in the Classification of Pigmented Skin Lesions |
| title_full | Deep Neural Frameworks Improve the Accuracy of General Practitioners in the Classification of Pigmented Skin Lesions |
| title_fullStr | Deep Neural Frameworks Improve the Accuracy of General Practitioners in the Classification of Pigmented Skin Lesions |
| title_full_unstemmed | Deep Neural Frameworks Improve the Accuracy of General Practitioners in the Classification of Pigmented Skin Lesions |
| title_short | Deep Neural Frameworks Improve the Accuracy of General Practitioners in the Classification of Pigmented Skin Lesions |
| title_sort | deep neural frameworks improve the accuracy of general practitioners in the classification of pigmented skin lesions |
| topic | artificial intelligence dermatology deep learning skin diseases melanoma |
| url | https://www.mdpi.com/2075-4418/10/11/969 |
| work_keys_str_mv | AT maximilianolucius deepneuralframeworksimprovetheaccuracyofgeneralpractitionersintheclassificationofpigmentedskinlesions AT jorgedeall deepneuralframeworksimprovetheaccuracyofgeneralpractitionersintheclassificationofpigmentedskinlesions AT joseantoniodeall deepneuralframeworksimprovetheaccuracyofgeneralpractitionersintheclassificationofpigmentedskinlesions AT martinbelvisi deepneuralframeworksimprovetheaccuracyofgeneralpractitionersintheclassificationofpigmentedskinlesions AT lucianaradizza deepneuralframeworksimprovetheaccuracyofgeneralpractitionersintheclassificationofpigmentedskinlesions AT marisalanfranconi deepneuralframeworksimprovetheaccuracyofgeneralpractitionersintheclassificationofpigmentedskinlesions AT victorialorenzatti deepneuralframeworksimprovetheaccuracyofgeneralpractitionersintheclassificationofpigmentedskinlesions AT carlosmgalmarini deepneuralframeworksimprovetheaccuracyofgeneralpractitionersintheclassificationofpigmentedskinlesions |