A Convolutional Neural Network
for Lentigo Diagnosis
Sana Zorgui
1(B) , Siwar Chaabene
1(B) , Bassem Bouaziz
1(B) , Hadj Batatia
2(B) ,
and Lotfi Chaari
2(B)
1 MIRACL and CRNS, University of Sfax, Sfax, Tunisia
sanazorgui@gmail.com, siwarchaabene@gmail.com,
bassem.bouaziz@isims.usf.tn
2 University of Toulouse, IRIT - INP-ENSEEIHT, Toulouse, France
{hadj.batatia,lotfi.chaari}@toulouse-inp.fr
Abstract. Using Reflectance Confocal Microscopy (RCM) for lentigo
diagnosis is today considered essential. Indeed, RCM allows fast data
acquisition with a high spatial resolution of the skin. In this paper, we
use a deep convolutional neural network (CNN) to perform RCM image
classification in order to detect lentigo. The proposed method relies on an
InceptionV3 architecture combined with data augmentation and transfer
learning. The method is validated on RCM data and shows very efficient
detection performance with more than 98% of accuracy.
Keywords: Reflectance Confocal Microscopy · Lentigo · CNN
classification · InceptionV3
1 Introduction
Reflectance Confocal Microscopy (RCM) [1] is a modality increasingly used in
medical imaging like MRI (Magnetic Resonance Imaging) [2–4] or X-ray imaging [5]. In vivo RCM technique is easy to use during the patient examination and
acquires high resolution skin images in a short time. This modality can be used
to help dermatologists diagnose different skin diseases. However, it takes a long
time for dermatologists to make full use of the possibilities of this technique for
diagnostic purposes. Our work aims to develop a new tool to automate certain
diagnostic steps required using deep learning [6]. On the other side, the lentigos
are age spots that mainly appear on the hand or on the areas most frequently
exposed to the sunlight. On the surface, they appear as a darker spot. Inside the
skin layers, it is mainly at the level of the dermis-epidermis junction that the
differences can be visible [7]. Therefore, the distinction of lentigos can be made
using the RCM images. Several deep learning architectures, especially convolutional neural network (CNN) [5,8] show great potential in medical imaging
classification. In this paper, we propose a new 3D RCM image (2D + depth)
classification method for lentigo detection. The method is based on a CNN on
InceptionV3 architecture [9].
c
The Author(s) 2020
M. Jmaiel et al. (Eds.): ICOST 2020, LNCS 12157, pp. 89–99, 2020.
https://doi.org/10.1007/978-3-030-51517-1_8
for Lentigo Diagnosis
Sana Zorgui
1(B) , Siwar Chaabene
1(B) , Bassem Bouaziz
1(B) , Hadj Batatia
2(B) ,
and Lotfi Chaari
2(B)
1 MIRACL and CRNS, University of Sfax, Sfax, Tunisia
sanazorgui@gmail.com, siwarchaabene@gmail.com,
bassem.bouaziz@isims.usf.tn
2 University of Toulouse, IRIT - INP-ENSEEIHT, Toulouse, France
{hadj.batatia,lotfi.chaari}@toulouse-inp.fr
Abstract. Using Reflectance Confocal Microscopy (RCM) for lentigo
diagnosis is today considered essential. Indeed, RCM allows fast data
acquisition with a high spatial resolution of the skin. In this paper, we
use a deep convolutional neural network (CNN) to perform RCM image
classification in order to detect lentigo. The proposed method relies on an
InceptionV3 architecture combined with data augmentation and transfer
learning. The method is validated on RCM data and shows very efficient
detection performance with more than 98% of accuracy.
Keywords: Reflectance Confocal Microscopy · Lentigo · CNN
classification · InceptionV3
1 Introduction
Reflectance Confocal Microscopy (RCM) [1] is a modality increasingly used in
medical imaging like MRI (Magnetic Resonance Imaging) [2–4] or X-ray imaging [5]. In vivo RCM technique is easy to use during the patient examination and
acquires high resolution skin images in a short time. This modality can be used
to help dermatologists diagnose different skin diseases. However, it takes a long
time for dermatologists to make full use of the possibilities of this technique for
diagnostic purposes. Our work aims to develop a new tool to automate certain
diagnostic steps required using deep learning [6]. On the other side, the lentigos
are age spots that mainly appear on the hand or on the areas most frequently
exposed to the sunlight. On the surface, they appear as a darker spot. Inside the
skin layers, it is mainly at the level of the dermis-epidermis junction that the
differences can be visible [7]. Therefore, the distinction of lentigos can be made
using the RCM images. Several deep learning architectures, especially convolutional neural network (CNN) [5,8] show great potential in medical imaging
classification. In this paper, we propose a new 3D RCM image (2D + depth)
classification method for lentigo detection. The method is based on a CNN on
InceptionV3 architecture [9].
c
The Author(s) 2020
M. Jmaiel et al. (Eds.): ICOST 2020, LNCS 12157, pp. 89–99, 2020.
https://doi.org/10.1007/978-3-030-51517-1_8