Automated detection, extraction and counting of acne lesions for automatic evaluation and tracking of acne severity

Maroni, Gabriele; Ermidoro, Michele; Previdi, Fabio; Bigini, Glauco

doi:10.1109/ssci.2017.8280925

Cited by 39 publications

(24 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A standardized, reproducible way to assess acne and their response to treatment would allow patients to self‐monitor through mobile applications, physicians to track progress remotely through tele‐dermatology, and research studies to be compared with each other in meta‐analyses . The use of digital image analysis has proven useful in the monitoring of several other skin conditions such as melasma and vitiligo, and deep learning has already had an impact in the automated diagnosis of skin cancers .…”

Section: Introductionmentioning

confidence: 99%

Automated grading of acne vulgaris by deep learning with convolutional neural networks

Lim

Akram

Ngo

et al. 2019

Skin Research and Technology

View full text Add to dashboard Cite

Background The visual assessment and severity grading of acne vulgaris by physicians can be subjective, resulting in inter‐ and intra‐observer variability. Objective To develop and validate an algorithm for the automated calculation of the Investigator's Global Assessment (IGA) scale, to standardize acne severity and outcome measurements. Materials and Methods A total of 472 photographs (retrieved 01/01/2004‐04/08/2017) in the frontal view from 416 acne patients were used for training and testing. Photographs were labeled according to the IGA scale in three groups of IGA clear/almost clear (0‐1), IGA mild (2), and IGA moderate to severe (3‐4). The classification model used a convolutional neural network, and models were separately trained on three image sizes. The photographs were then subjected to analysis by the algorithm, and the generated automated IGA scores were compared to clinical scoring. The prediction accuracy of each IGA grade label and the agreement (Pearson correlation) of the two scores were computed. Results The best classification accuracy was 67%. Pearson correlation between machine‐predicted score and human labels (clinical scoring and researcher scoring) for each model and various image input sizes was 0.77. Correlation of predictions with clinical scores was highest when using Inception v4 on the largest image size of 1200 × 1600. Two sets of human labels showed a high correlation of 0.77, verifying the repeatability of the ground truth labels. Confusion matrices show that the models performed sub‐optimally on the IGA 2 label. Conclusion Deep learning techniques harnessing high‐resolution images and large datasets will continue to improve, demonstrating growing potential for automated clinical image analysis and grading.

show abstract

Section: Introductionmentioning

confidence: 99%

Automated grading of acne vulgaris by deep learning with convolutional neural networks

Lim

Akram

Ngo

et al. 2019

Skin Research and Technology

View full text Add to dashboard Cite

show abstract

“…To conduct a systematic review on acne images segmentation methods, an adapted PRISMA (Preferred Reporting Items for Systematic reviews and Meta-Analyses) (Liberati et al, 2009) standard was used. The analysis includes all studies published until April 2019 and explores four databases: Scopus (https://www.scopus.com/home.uri), PubMed Central The review shows that current segmentation methods for acne vulgaris images can be divided into two groups: those algorithms based on classical image processing techniques (Ramli, Malik, Hani, & Yap, 2011a;Chen, Chang, & Cao, 2012;Khongsuwan, Kiattisin, Wongseree, & Leelasantitham, 2012;Humayun, Malik, Belhaouari, Kamel, & Yap, 2012;Liu & Zerubia, 2013;Min, Kong, Yoon, Kim, & Suh, 2013;Malik, Humayun, Kamel, & Yap, 2014;Chantharaphaichi, Uyyanonvara, Sinthanayothin, & Nishihara, 2015;Alamdari, Tavakolian, Alhashim, & Fazel-Rezai, 2016;Kittigul & Uyyanonvara, 2016;Budhi, Adipranata, & Gunawan, 2017;Maroni, Ermidoro, Previdi, & Bigini, 2017) -they consist of a series of steps or operations that have to be applied to an image, for instance color space transformations or contrast modifications. The other group refers to machine learning algorithms (Fujii et al, 2008;Ramli, Malik, Hani, & Yap, 2011b;Madan, Dana, & Cula, 2011;Arifin, Kibria, Firoze, Amini, & Yan, 2012;Chang & Liao, 2013;Khan, Malik, Kamel, Dass, & Affandi, 2015;Alamdari et al, 2016).…”

Section: Systematic Reviewmentioning

confidence: 99%

“…An analysis of limitations for each study included in the systematic review showed that algorithms based on classical image processing techniques cannot be totally automatized, mainly because there are some parameters that need to be manually adjusted (Son et al, 2008;Humayun et al, 2012;Budhi et al, 2017;Maroni et al, 2017). That is why in the present work machine learning algorithms are chosen for the implementation of the proposed methodology.…”

Section: Systematic Reviewmentioning

confidence: 99%

Analysis of segmentation methods for acne vulgaris images. Proposal of a new methodology applied to fluorescence images

Perís-Fajarnés

Santonja

Alamán

et al. 2019

Proceedings INNODOCT/19. International Conference on Innovation, Documentation and Education

View full text Add to dashboard Cite

Acne vulgaris is one of the most common human pathologies worldwide. Its prevalence causes a high healthcare expenditure. Acne healthcare costs and effects on individuals' quality of life lead to the need of analysing current acne evaluation, treatment and monitoring methods. One of the most common ones is manual lesion counting by a dermatologist. However, this technique has several limitations, such as time spent. That is the reason why the development of new computer-assisted techniques are needed in order to automatically count the acne lesions. Nonetheless, the first step is automatic acne lesion detection on the skin of patients. The aim of this work is to propose a new methodology to solve the acne images segmentation problem, so that the implementation of a system for automatic counting is possible. The results would be a decrease in both time spent and diagnosis errors. With this objective, after doing a systematic review on the state of the art of acne images segmentation methods, fluorescence images of the face of acne patients are obtained. This image modality enhances visualization of the acne lesions. Finally, using the fluorescence images, a segmentation algorithm is implemented in MATLAB.

show abstract

“…The existing methodologies for acne fluorescence images are both based on classical image processing techniques. However, from the analysis of limitations declared on each analysed study, it can be concluded that algorithms based on these techniques cannot be totally automatized, mainly because there are some parameters that need to be manually adjusted 13,20‐22 . That is why in the present work machine learning algorithms are chosen for the implementation of the proposed methodology.…”

Section: Introductionmentioning

confidence: 99%

Segmentation methods for acne vulgaris images: Proposal of a new methodology applied to fluorescence images

Perís-Fajarnés

Santonja

García

et al. 2020

Skin Research and Technology

View full text Add to dashboard Cite

Acne vulgaris is an inflammatory chronic disease of pilosebaceous units-a pilosebaceous unit is formed by all the hair follicles related to the same sebaceous gland. The main affected regions are face, neck, chest and back. Clinical manifestations are seborrhoea or overproduction of sebum, the presence of non-inflammatory and inflammatory lesions on skin (open and closed commedos, and pustules and papules, respectively), and scars. 1 It is the most common dermatological pathology worldwide. 2 85% of adolescents suffer from acne. 3 In order to monitor and treat acne properly, a precise and reliable method to establish acne severity is needed. 4 Currently, there is a wide range of acne grading systems, which shows the lack of a global standard. These systems can be divided into two groups: those based on manual lesion counting and the ones that use a model photography. 3,4 Several studies have proven lesion counting to be an objective and reliable method. 4,5

show abstract

Automated detection, extraction and counting of acne lesions for automatic evaluation and tracking of acne severity

Cited by 39 publications

References 13 publications

Automated grading of acne vulgaris by deep learning with convolutional neural networks

Automated grading of acne vulgaris by deep learning with convolutional neural networks

Analysis of segmentation methods for acne vulgaris images. Proposal of a new methodology applied to fluorescence images

Segmentation methods for acne vulgaris images: Proposal of a new methodology applied to fluorescence images

Contact Info

Product

Resources

About