Romain Cendre

Bio: Romain Cendre is an academic researcher from French Institute of Health and Medical Research. The author has contributed to research in topics: Feature extraction & Likert scale. The author has an hindex of 3, co-authored 7 publications receiving 24 citations.

Radiology resident MR and CT image analysis skill assessment using an interactive volumetric simulation tool – the RadioLOG project

Abstract: Assess the use of a volumetric simulation tool for the evaluation of radiology resident MR and CT interpretation skills. Forty-three participants were evaluated with a software allowing the visualisation of multiple volumetric image series. There were 7 medical students, 28 residents and 8 senior radiologists among the participants. Residents were divided into two sub-groups (novice and advanced). The test was composed of 15 exercises on general radiology and lasted 45 min. Participants answered a questionnaire on their experience with the test using a 5-point Likert scale. This study was approved by the dean of the medical school and did not require ethics committee approval. The reliability of the test was good with a Cronbach alpha value of 0.9. Test scores were significantly different in all sub-groups studies (p < 0.0225). The relation between test scores and the year of residency was logarithmic (R2 = 0.974). Participants agreed that the test reflected their radiological practice (3.9 ± 0.9 on a 5-point scale) and was better than the conventional evaluation methods (4.6 ± 0.5 on a 5-point scale). This software provides a high quality evaluation tool for the assessment of the interpretation skills in radiology residents. • This tool allows volumetric image analysis of MR and CT studies. • A high reliability test could be created with this tool. • Test scores were strongly associated with the examinee expertise level. • Examinees positively evaluated the authenticity and usability of this tool.

Two Schemes for Automated Diagnosis of Lentigo on Confocal Microscopy Images

Abstract: Reflectance Confocal Microscopy is an imaging modality increasingly used for diagnosis of skin pathologies in clinical context thanks to specific and rich information they provide. However, few studies apply automatic methods for prediction in this kind of images. In this paper, we investigate in this paper a classification on these images on three categories: Healthy, Benign and Malignant Lentigo. To this end, we implement three feature extraction methods, namely Wavelets, Haralick and CNN through Transfer Learning. Furthermore, we exploit these feature extraction within two approaches: the first one operates on the entire image and the second one operates at patch-level (multiple patches per image) by giving a score to each patch. The scores are merged later to build a final decision for an image. Results show that Transfer learning obtains the best results for the two approaches, particularly with Average pooling.

Classification of Lentigo Maligna at Patient-Level by Means of Reflectance Confocal Microscopy Data

Abstract: Featured Application: This paper focuses on improvement in patient care and it also helps practitioners optimize their dermatology services by means of computer-assisted diagnostic software using data from reflectance confocal microscopy devices. Abstract: Reflectance confocal microscopy is an appropriate tool for the diagnosis of lentigo maligna. Compared with dermoscopy, this device can provide abundant information as a mosaic and/or a stack of images. In this particular context, the number of images per patient varied between 2 and 833 images and the objective, ultimately, is to be able to discern between benign and malignant classes. First, this paper evaluated classification at the image level, with the help of handcrafted methods derived from the literature and transfer learning methods. The transfer learning feature extraction methods outperformed the handcrafted feature extraction methods from literature, with a F 1 score value of 0.82. Secondly, this work proposed patient-level supervised methods based on image decisions and a comparison of these with multi-instance learning methods. This study achieved comparable results to those of the dermatologists, with an AUC score of 0.87 for supervised patient diagnosis and an AUC score of 0.88 for multi-instance learning patient diagnosis. According to these results, computer-aided diagnosis methods presented in this paper could be easily used in a clinical context to save time or confirm a diagnosis and can be oriented to detect images of interest. Also, this methodology can be used to serve future works based on multimodality.

Deep Learning in Biomedical Optics

Abstract: This article reviews deep learning applications in biomedical optics with a particular emphasis on image formation. The review is organized by imaging domains within biomedical optics and includes microscopy, fluorescence lifetime imaging, in vivo microscopy, widefield endoscopy, optical coherence tomography, photoacoustic imaging, diffuse tomography, and functional optical brain imaging. For each of these domains, we summarize how deep learning has been applied and highlight methods by which deep learning can enable new capabilities for optics in medicine. Challenges and opportunities to improve translation and adoption of deep learning in biomedical optics are also summarized. Lasers Surg. Med. © 2021 Wiley Periodicals LLC.

The Impact of Digital Technologies on Public Health in Developed and Developing Countries : 18th International Conference, ICOST 2020, Hammamet, Tunisia, June 24–26, 2020, Proceedings

Abstract: The objective of this work is to detect Alzheimer’s disease using Magnetic Resonance Imaging. For this, we use a three-dimensional densenet-121 architecture. With the use of only freely available tools, we obtain good results: a deep neural network showing metrics of 87% accuracy, 87% sensitivity (micro-average), 88% specificity (micro-average), and 92% AUROC (micro-average) for the task of classifying five different classes (disease stages). The use of tools available for free means that this work can be replicated in developing countries.

A Convolutional Neural Network for Lentigo Diagnosis

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.

Integrating Growth Variability of the Ilium, Fifth Lumbar Vertebra, and Clavicle with Multivariate Adaptive Regression Splines Models for Subadult Age Estimation.

Abstract: Subadult age estimation should rely on sampling and statistical protocols capturing development variability for more accurate age estimates. In this perspective, measurements were taken on the fifth lumbar vertebrae and/or clavicles of 534 French males and females aged 0-19 years and the ilia of 244 males and females aged 0-12 years. These variables were fitted in nonparametric multivariate adaptive regression splines (MARS) models with 95% prediction intervals (PIs) of age. The models were tested on two independent samples from Marseille and the Luis Lopes reference collection from Lisbon. Models using ilium width and module, maximum clavicle length, and lateral vertebral body heights were more than 92% accurate. Precision was lower for postpubertal individuals. Integrating punctual nonlinearities of the relationship between age and the variables and dynamic prediction intervals incorporated the normal increase in interindividual growth variability (heteroscedasticity of variance) with age for more biologically accurate predictions.