Browsing by Author "Navarrete-Dechent, Cristian"

Now showing 1 - 20 of 37
  • Thumbnail Image
    Item
    A systematic review and meta-analysis of artificial intelligence versus clinicians for skin cancer diagnosis
    (2024) Salinas, María Paz; Sepúlveda, Javiera; Hidalgo, Leonel; Peirano, Dominga; Morel, Macarena; Uribe, Pablo; Rotemberg, Verónica; Briones, Juan; Mery, Domingo; Navarrete-Dechent, Cristian
    Scientific research of artificial intelligence (AI) in dermatology has increased exponentially. The objective of this study was to perform a systematic review and meta-analysis to evaluate the performance of AI algorithms for skin cancer classification in comparison to clinicians with different levels of expertise. Based on PRISMA guidelines, 3 electronic databases (PubMed, Embase, and Cochrane Library) were screened for relevant articles up to August 2022. The quality of the studies was assessed using QUADAS-2. A meta-analysis of sensitivity and specificity was performed for the accuracy of AI and clinicians. Fifty-three studies were included in the systematic review, and 19 met the inclusion criteria for the meta-analysis. Considering all studies and all subgroups of clinicians, we found a sensitivity (Sn) and specificity (Sp) of 87.0% and 77.1% for AI algorithms, respectively, and a Sn of 79.78% and Sp of 73.6% for all clinicians (overall); differences were statistically significant for both Sn and Sp. The difference between AI performance (Sn 92.5%, Sp 66.5%) vs. generalists (Sn 64.6%, Sp 72.8%), was greater, when compared with expert clinicians. Performance between AI algorithms (Sn 86.3%, Sp 78.4%) vs expert dermatologists (Sn 84.2%, Sp 74.4%) was clinically comparable. Limitations of AI algorithms in clinical practice should be considered, and future studies should focus on real-world settings, and towards AI-assistance.
  • No Thumbnail Available
    Item
    Accuracy in anatomical location on dermatological surgery: a multi‐centre retrospective study
    (2023) Donoso, Francisca; Hidalgo, Leonel; Cowen, Emily A.; Villagran, Sofía; Villablanca, Paula; Puerto, Constanza del; Silva‐Valenzuela, Sergio; Galimany, Lucas; Majerson, Daniela; Andino, Romina; Uribe, Pablo; Droppelmann, Katherine; Cárdenas, Consuelo; Abarzúa‐Araya, Álvaro; Castro‐Ayala, Juan C.; Kurtansky, Nicholas R.; Halpern, Allan C.; Molenda, Matthew A.; Rotemberg, Veronica; Navarrete-Dechent, Cristian
  • Thumbnail Image
    Item
    Assessing field cancerization and actinic keratosis using ultraviolet-induced fluorescence dermatoscopy after the application of 5-aminolevulinic acid - An observational study
    (2024) Korecka, Katarzyna; Polanska, Adriana; Danczak-Pazdrowska, Aleksandra; Navarrete-Dechent, Cristian
    Background: Actinic keratoses (AK) are one of the most frequent reasons for consultations in dermatology. Ultraviolet-induced fluorescence dermatoscopy (UVFD) is a new method that allows the assessment of lesions in a spectrum of light that originates from the fluorochromes emitting UV-excited luminescence. The aim of this study was to assess the UVFD features of AKs before PDT and their intensity in field cancerization and single lesions. Methods: This retrospective study was conducted from June to November 2023. Lesions were assessed with the Olsen scale clinically and dermatoscopically (DermLite DL5, 10x magnification) and photographed. UVFD fluorescence was categorized as 'none', 'weak', 'moderate', and 'intense'. A 1-mm thick layer of 10 % 5-ALA gel was applied to single lesions or cancerization field (depending on the patient) and covered with an occlusive dressing for 3 h. Prior the application of 10 % 5-ALA gel, the lesions were degreased with an alcoholic solution. The occlusion was removed, and the field was cleaned with a 0,9 % saline solution. Afterward, each lesion was photographed in polarized light and UVFD mode. Results: A total of 194 dermatoscopic images were analyzed, 111 corresponded to field cancerization and 81 to single AKs. Overall, weak fluorescence was noticed in 22 of them (11,3 %), moderate in 107 (55,15 %), and intense in 65 (33,5 %). Amongst field cancerization (111 images), weak fluorescence was seen in 11 (9.9 %), moderate in 68 (61,26 %), and intense in 32 (28,82 %). In single lesions (81 images), weak fluorescence was detected in 11 (13,2 %), moderate in 39 (46,99 %), and intense in 33 (28.83 %) of the lesions. Slightly more intense fluorescence was noticed in higher Olsen grade (p = 0.04). Conclusions: UVFD can enhance our efficacy of pre-procedural examination and might arise as a useful device to predict the therapeutic effect of PDT.
  • No Thumbnail Available
    Item
    Bilateral diffuse uveal melanocytic proliferation with multifocal diffuse integumentary melanocytic proliferation paraneoplastic syndrome: A case report
    (2021) Navarrete-Dechent, Cristian; Monnier, Jilliana; Marghoob, Nadeem G.; Liopyris, Konstantinos; Busam, Klaus J.; Francis, Jasmine H.; Marghoob, Ashfaq A.
    Bilateral diffuse uveal melanocytic proliferation (B-DUMP) is a rare paraneoplastic syndrome typically presenting with bilateral visual loss. B-DUMP is associated with extraocular systemic malignancies with the most common being lung cancer in males and uro-gynaecological cancer in females (mainly ovarian cancer). Cutaneous and/or mucosal involvement in patients with B-DUMP has been reported but it is not well characterised. Herein, we present a female in her 70s with diagnosis of stage IV vaginal clear-cell carcinoma and metastatic melanoma of unknown primary that developed progressive bilateral loss of visual acuity compatible with `B-DUMP'. Simultaneously, she developed multifocal bilateral bluish-greyish patches on the skin that were shown to have a proliferation of dermal melanocytes. We propose that the clinical and histopathologic cutaneous findings seen in patients with B-DUMP be termed 'diffuse integumentary melanocytic proliferation (DIMP)'.
  • Thumbnail Image
    Item
    Biopsy type does not affect the number of stages during Mohs micrographic surgery: a retrospective study
    (Oxford University Press, 2023) Araneda Ortega, Paulina Belén; Donoso Mena, Francisca; Castro, Juan C.; Uribe González, Pablo Francisco; Rossi, Anthony M.; Hibler, Brian P.; Droppelmann Droppelmann, Katherine Ann; Cárdenas De La Torre, Consuelo Paz; Navarrete-Dechent, Cristian
    Mohs micrographic surgery (MMS) is the treatment of choice for high-risk basal cell carcinoma (BCC). However, there are no evidence-based recommendations regarding which biopsy type is more appropriate to obtain tumour samples prior to MMS. Shave or punch biopsies are performed depending on the clinical characteristics of the tumour, surgeon experience and local protocols. However, biopsy type might result in difficult histopathological interpretation and influence the practical implementation of MMS. We performed a retrospective study on 208 consecutive BCCs treated with MMS. Of the 208 BCC biopsies, 42 (20.2%) were obtained by the shave method and 166 (79.8%) via punch. Those obtained with the shave technique had a mean of 1.64 stages vs. 1.69 stages with the punch technique (P = 0.130). These findings suggest biopsy type does not affect Mohs surgery performance. The biopsy type of choice is the one deemed adequate for each specific case to obtain a diagnosis and tumour subtyping., In this study including 208 primary basal cell carcinomas undergoing Mohs micrographic surgery (MMS), there were no differences in the mean number of stages, regardless of which biopsy type was performed (shave vs. punch). The biopsy technique might affect correct subtype identification; however, there were also no differences in the rate of upstaging. Any tissue reaction could also result in difficult histopathological interpretation on frozen sections; however, this was not evident in our study. It seems that biopsy type, shave vs. punch, does not affect MMS performance; thus the more appropriate biopsy type is the one deemed adequate for each specific case in order to obtain a diagnosis as well as tumour subtyping.
  • No Thumbnail Available
    Item
    Clinical size is a poor predictor of invasion in melanoma of the lentigo maligna type
    (2021) Navarrete-Dechent, Cristian; Aleissa, Saud; Connolly, Karen; Hibler, Brian P.; Dusza, Stephen W.; Rossi, Anthony M.; Lee, Erica; Nehal, Kishwer S.
    Background: There are no well-defined clinical factors to predict the risk of occult invasion in melanoma of the lentigo maligna type (LM) before complete histopathologic analysis.
  • No Thumbnail Available
    Item
    Contemporary management of actinic keratosis
    (2021) Navarrete-Dechent, Cristian; Marghoob, Ashfaq A.; Marchetti, Michael A.
    Actinic keratosis (AK) is a skin lesion characterized by itraepithelial keratinocyte dysplasia and molecular alterations shared with normal chronically sun-damaged skin and squamous cell carcinoma (SCC). AK can undergo spontaneous regression, stable existence, or malignant transformation to cutaneous SCC with progression rates to SCC ranging from 0% to 0.5% per lesion-year and AK spontaneous regression of 15-63%. As AK is a potential precursor of invasive SCC, it is commonly treated to mitigate the risk of malignant progression, including metastasis and death. There is a myriad of available spots (e.g. cryotherapy) and field (e.g. 5-fluorouracil, imiquimod photodynamic therapy) treatments for AK. Recently published randomized clinical trials have helped bridge the gap on AK management. In this viewpoint, we sought to summarize the most up-to-date evidence in the management of AK.
  • No Thumbnail Available
    Item
    Deep Learning for Basal Cell Carcinoma Detection for Reflectance Confocal Microscopy
    (2022) Campanella, Gabriele; Navarrete-Dechent, Cristian; Liopyris, Konstantinos; Monnier, Jilliana; Aleissa, Saud; Minhas, Brahmteg; Scope, Alon; Longo, Caterina; Guitera, Pascale; Pellacani, Giovanni; Kose, Kivanc; Halpern, Allan C.; Fuchs, Thomas J.; Jain, Manu
    Basal cell carcinoma (BCC) is the most common skin cancer, with over 2 million cases diagnosed annually in the UnitedStates. Conventionally, BCC is diagnosed by naked eye examination and dermoscopy. Suspicious lesions are either removed or biopsied for histopathological confirmation, thus lowering the specificity of noninvasive BCC diagnosis. Recently, reflectance confocal microscopy, a noninvasive diagnostic technique that can image skin lesions at cellular level resolution, has shown to improve specificity in BCC diagnosis and reduced the number needed to biopsy by 2-3 times. In this study, we developed and evaluated a deep learning-based artificial intelligence model to automatically detect BCC in reflectance confocal microscopy images. The proposed model achieved an area under the curve for the receiver operator characteristic curve of 89.7%(stack level) and 88.3%(lesion level), a performance on par with that of reflectance confocal microscopy experts. Furthermore, themodel achieved an area under the curve of 86.1% on a held-out test set from international collaborators, demonstrating the reproducibility and generalizability of the proposed automated diagnostic approach. These results provide a clear indication that the clinical deployment of decision support systems for the detection of BCC in reflectance confocal microscopy images has the potential for optimizing the evaluation and diagnosis of patients with skin cancer.
  • No Thumbnail Available
    Item
    Delphi Consensus Among International Experts on the Diagnosis, Management, and Surveillance for Lentigo Maligna
    (2023) Longo, Caterina; Navarrete-Dechent, Cristian; Tschandl, Philipp; Apalla, Zoe; Argenziano, Giuseppe; Braun, Ralph P.; Bataille, Veronique; Cabo, Horacio; Hoffmann-Wellhenhof, Rainer; Forsea, Ana Maria; Garbe, Claus; Guitera, Pascale; Raimond, Karls; Marghoob, Ashfaq A.; Malvehy, Josep; Del Marmol, Veronique; Moreno, David; Nehal, Kishwer S.; Nagore, Eduardo; Paoli, John; Pellacani, Giovanni; Peris, Ketty; Puig, Susana; Soyer, H. Peter; Swetter, Susan; Stratigos, Alexander; Stolz, Wilhelm; Thomas, Luc; Tiodorovic, Danica; Zalaudek, Iris; Kittler, Harald; Lallas, Aimilios
    Introduction: Melanoma of the lentigo maligna (LM) type is challenging. There is lack of consensus on the optimal diagnosis, treatment, and follow-up.
  • Thumbnail Image
    Item
    Dermatologist-like explainable AI enhances trust and confidence in diagnosing melanoma
    (2024) Chanda, Tirtha; Hauser, Katja; Hobelsberger, Sarah; Bucher, Tabea-Clara; Garcia, Carina Nogueira; Wies, Christoph; Kittler, Harald; Tschandl, Philipp; Navarrete-Dechent, Cristian; Podlipnik, Sebastian; Chousakos, Emmanouil; Crnaric, Iva; Majstorovic, Jovana; Alhajwan, Linda; Foreman, Tanya; Peternel, Sandra; Sarap, Sergei; Ozdemir, Irem; Barnhill, Raymond L.; Llamas-Velasco, Mar; Poch, Gabriela; Korsing, Soeren; Sondermann, Wiebke; Gellrich, Frank Friedrich; Heppt, Markus V.; Erdmann, Michael; Haferkamp, Sebastian; Drexler, Konstantin; Goebeler, Matthias; Schilling, Bastian; Utikal, Jochen S.; Ghoreschi, Kamran; Froehling, Stefan; Krieghoff-Henning, Eva; Brinker, Titus J.
    Artificial intelligence (AI) systems have been shown to help dermatologists diagnose melanoma more accurately, however they lack transparency, hindering user acceptance. Explainable AI (XAI) methods can help to increase transparency, yet often lack precise, domain-specific explanations. Moreover, the impact of XAI methods on dermatologists' decisions has not yet been evaluated. Building upon previous research, we introduce an XAI system that provides precise and domain-specific explanations alongside its differential diagnoses of melanomas and nevi. Through a three-phase study, we assess its impact on dermatologists' diagnostic accuracy, diagnostic confidence, and trust in the XAI-support. Our results show strong alignment between XAI and dermatologist explanations. We also show that dermatologists' confidence in their diagnoses, and their trust in the support system significantly increase with XAI compared to conventional AI. This study highlights dermatologists' willingness to adopt such XAI systems, promoting future use in the clinic.
  • Thumbnail Image
    Item
    Dermoscopic Features of Pigmented Bowen Disease: A Multicenter Study on Behalf of the Ibero-Latin American College of Dermatology (CILAD)
    (2024) Cabo, Horacio; Salerni, Gabriel; Sabban, Emilia Cohen; Garlatti, Agustin Bollea; Orendain, Nicole; Rodriguez-Saa, Sonia; Bakos, Renato Marchiori; Pozzobon, Flavia Carolina; Gonzalez, Virginia M.; Peralta, Rosario; Navarrete-Dechent, Cristian; Peirano, Dominga; Perez-Fernandez, Elia; Puig, Susana
    Introduction: Studies focused on dermoscopic aspects of pigmented Bowen disease (pBD) in Latin American population are scarce and limited to only case reports or small series. Objectives: To report dermoscopic findings in a large series of 147 pBD diagnosed in Ibero-Latin American population. Methods: We conducted a multicentric, retrospective study on 147 histologically proven pBD under the auspices of the Dermoscopy Chapter of the Ibero-Latin American College of Dermatology. Results: The study population consisted of 77 females (52%) and 70 males (48%) with a mean age of 68.6 years. 70.1% of patients had skin phototype 3, 15.6% to skin phototype 2, and 14.3% to skin phototype 4. On clinical examination, near 60% of pBD were flat, 70% presented with scales, and 90% were asymmetric. Under dermoscopy, structureless hypopigmented areas, dots brown and pink color were the most frequently observed. Regarding specific dermoscopic clues to pBD, the most prevalent were structureless hypopigmented areas, vessels arranged in linear fashion at the periphery, and pigmented lines or pigmented dots distributed in a linear fashion. Clustered, coiled, and dotted vessels were observed in 55.8%, 45.6%, and 45.6% of the cases, respectively. Conclusions: We report a large series of cases of pBD in Latin American patients, with most patients being skin phototype 3 and 4. Distinctively in our study, the pigmented structures and the clues derived from the presence of melanin were much more frequent than in previous reports in fair skin.
  • No Thumbnail Available
    Item
    Dermoscopy as a clinical tool for the diagnosis of demodicosis: a retrospective intrapatient case-control study
    (2024) Parra-Cares, Julio; Meza-Romero, Rodrigo; Ibanez, Samuel; Canales, Marilena; Concha, Monica; Navarrete-Dechent, Cristian; Abarzua-Araya, Alvaro
    Dermoscopy has been used for the non-invasive diagnosis of demodicosis. Several studies have evaluated the usefulness of this tool in the diagnosis, however, there are differences in the gold standard (SSSB or KOH test) and criteria of positivity used between studies. Added to this, is the lack of controls and objective quantification of the usefulness of dermoscopic signs in clinically observable and relevant ranges. To validate the usefulness of dermoscopy for the diagnosis of demodicosis by calculating the performance indicators for the different dermoscopic signs. Retrospective intrapatient case-control study, which included adults with suspicion of demodicosis. Dermoscopic photographs and scraping of healthy and lesional skin were obtained. Samples were analyzed microscopically by trained personnel. Photographs were evaluated by determining the presence of Demodex tails (DT), dilated follicular openings (DFO) and dilated blood vessels (DBV) in pre-defined ranges. 64 patients were included (total = 256 samples); the presence of demodex on skin scraping was seen in 69%. Under dermoscopy, the presence of DT in range 11-20/field had a positive likelihood ratio (LR) of 12.10 (95%CI 6.52-22.45) and negative LR 0.32 (95%CI 0.23-0.45). Combined and dichotomized performance for at least one positive sign under dermoscopy (DT > 10/field, DFO > 10/field or DBV > 50% of the field): positive LR 7.14 (95%CI 4.80-10.62) and negative LR 0.11 (95%CI 0.06-0.22). The presence of DT, DFO or DBV has a high correlation with a positive mite test, so the diagnosis of demodicosis could be made only through dermoscopy.
  • No Thumbnail Available
    Item
    Differential expression of programmed cell death ligand 1 (PD-L1) and inflammatory cells in basal cell carcinoma subtypes
    (2022) Gompertz-Mattar, Matias; Perales, Juan; Sahu, Aditi; Mondaca, Sebastian; Gonzalez, Sergio; Uribe, Pablo; Navarrete-Dechent, Cristian
    Few studies have evaluated programmed cell death ligand (PD-L1) expression and lymphocytic infiltrates in Basal Cell Carcinoma (BCC). The objectives of this study are to assess PD-L1 expression and markers of local immune response in nodular, superficial, and morpheaform BCC, and compare it to normal, sun-exposed skin from the periphery of intradermal nevi. This was a retrospective study that included three histological subtypes of BCCs, and sun-exposed skin from the periphery of dermal nevi as quality controls. Tissue microarrays (TMA) were constructed with subsequent staining of H&E and immunohistochemistry (IHC) for CD4, CD8, FOXP3 and PD-L1. Non-automated quantification of the infiltrate in the intratumoral and stromal compartments on TMAs was performed. A total of 115 BCC (39 nodular, 39 morpheaform, and 37 superficial) and 41 sun-exposed skin samples were included (mean age 65.4 years; 52.6% females). BCC showed higher expression of PD-L1 (5.4 vs 0.7%, p < 0.001), CD8 (29.8 vs 19.7%, p = 0.002), and FOXP3 (0.3 vs 0.06%, p = 0.022) compared to sun-exposed skin. There was a higher PD-L1 expression in nodular BCC compared with other subtypes. Low-risk BCC subtypes (superficial and nodular) exhibited more PD-L1 expression in intratumoral and stromal immune infiltrates as compared to high-risk BCC subtypes. As a limitation, no immune cells function was evaluated in this study, only the presence/absence of T-lymphocyte sub-populations was recorded. Substantial differences in both PD-L1 expression and lymphocytic infiltrates were found amongst the histological subtypes of BCC and sun-exposed skin. Highest PD-L1 expression was found in nodular BCCs which suggests a potentially targetable strategy in the treatment of this most common BCC subtype.
  • No Thumbnail Available
    Item
    Differentiating Fordyce Spots from Their Common Simulators Using Ultraviolet-Induced Fluorescence Dermatoscopy-Retrospective Study
    (2023) Pietkiewicz, Pawel; Navarrete-Dechent, Cristian; Goldust, Mohamad; Korecka, Katarzyna; Todorovska, Verce; Errichetti, Enzo
    Fordyce spots (FS) are heterotopic sebaceous glands affecting mostly oral and genital mucosa, commonly misdiagnosed with sexually transmitted infections. In a single-center retrospective study, we aimed to assess the ultraviolet-induced fluorescencedermatoscopy (UVFD) clues of Fordyce spots and their common clinical simulants: molluscum contagiosum, penile pearly papules, human papillomavirus warts, genital lichen planus, and genital porokeratosis. Analyzed documentation included patients' medical records (1 September-30 October 2022) and photodocumentation, which included clinical images as well as polarized, non-polarized, and UVFD images. Twelve FS patients were included in the study group and fourteen patients in the control group. A novel and seemingly specific UVFD pattern of FS was described: regularly distributed bright dots over yellowish-greenish clods. Even though, in the majority of instances, the diagnosis of FS does not require more than naked eye examination, UVFD is a fast, easy-to-apply, and low-cost modality that can further increase the diagnostic confidence and rule out selected infectious and non-infectious differential diagnoses if added to conventional dermatoscopic diagnosis.
  • No Thumbnail Available
    Item
    Einfache Hamostase an der Kopfhaut bei dermatologischen Operationen durch Kochsalzlosung
    (2023) Hidalgo, Leonel; Carrasco, Karina; Navarrete-Dechent, Cristian; Chen, Curtis; Uribe, Pablo; Carreno, Nestor
  • No Thumbnail Available
    Item
    Evaluation of Artificial Intelligence-Assisted Diagnosis of Skin Neoplasms: A Single-Center, Paralleled, Unmasked, Randomized Controlled Trial
    (2022) Han, Seung Seog; Kim, Young Jae; Moon, Ik Jun; Jung, Joon Min; Lee, Mi Young; Lee, Woo Jin; Won, Chong Hyun; Lee, Mi Woo; Kim, Seong Hwan; Navarrete-Dechent, Cristian; Chang, Sung Eun
    Trial design: This was a single-center, unmasked, paralleled, randomized controlled trial. Methods: A randomized trial was conducted in a tertiary care institute in South Korea to validate whether artificial intelligence (AI) could augment the accuracy of nonexpert physicians in the real-world settings, which included diverse outof-distribution conditions. Consecutive patients aged >19 years, having one or more skin lesions suspicious for skin cancer detected by either the patient or physician, were randomly allocated to four nondermatology trainees and four dermatology residents. The attending dermatologists examined the randomly allocated patients with (AI-assisted group) or without (unaided group) the real-time assistance of AI algorithm (https:// b2020.modelderm.com#world; convolutional neural networks; unmasked design) after simple randomization of the patients. Results: Using 576 consecutive cases (Fitzpatrick skin phototypes III or IV) with suspicious lesions out of the initial 603 recruitments, the accuracy of the AI-assisted group (n = 295, 53.9%) was found to be significantly higher than those of the unaided group (n = 281, 43.8%; P = 0.019). Whereas the augmentation was more significant from 54.7% (n = 150) to 30.7% (n = 138; P < 0.0001) in the nondermatology trainees who had the least experience in dermatology, it was not significant in the dermatology residents. The algorithm could help trainees in the AI-assisted group include more differential diagnoses than the unaided group (2.09 vs. 1.95 diagnoses; P = 0.0005). However, a 12.2% drop in Top-1 accuracy of the trainees was observed in cases in which all Top-3 predictions given by the algorithm were incorrect. Conclusions: The multiclass AI algorithm augmented the diagnostic accuracy of nonexpert physicians in dermatology.
  • No Thumbnail Available
    Item
    Expert Agreement on the Presence and Spatial Localization of Melanocytic Features in Dermoscopy
    (2024) Liopyris, Konstantinos; Navarrete-Dechent, Cristian; Marchetti, Michael A.; Rotemberg, Veronica; Apalla, Zoe; Argenziano, Giuseppe; Blum, Andreas; Braun, Ralph P.; Carrera, Cristina; Codella, Noel C. F.; Combalia, Marc; Dusza, Stephen W.; Gutman, David A.; Helba, Brian; Hofmann-Wellenhof, Rainer; Jaimes, Natalia; Kittler, Harald; Kose, Kivanc; Lallas, Aimilios; Longo, Caterina; Malvehy, Josep; Menzies, Scott; Nelson, Kelly C.; Paoli, John; Puig, Susana; Rabinovitz, Harold S.; Rishpon, Ayelet; Russo, Teresa; Scope, Alon; Soyer, H. Peter; Stein, Jennifer A.; Stolz, Willhelm; Sgouros, Dimitrios; Stratigos, Alexander J.; Swanson, David L.; Thomas, Luc; Tschandl, Philipp; Zalaudek, Iris; Weber, Jochen; Halpern, Allan C.; Marghoob, Ashfaq A.
    Dermoscopy aids in melanoma detection; however, agreement on dermoscopic features, including those of high clinical relevance, remains poor. In this study, we attempted to evaluate agreement among experts on exemplar images not only for the presence of melanocytic-specific features but also for spatial localization. This was a cross-sectional, multicenter, observational study. Dermoscopy images exhibiting at least 1 of 31 melanocytic-specific features were submitted by 25 world experts as exemplars. Using a web-based platform that allows for image markup of specific contrast-defined regions (superpixels), 20 expert readers annotated 248 dermoscopic images in collections of 62 images. Each collection was reviewed by five independent readers. A total of 4,507 feature observations were performed. Good-to-excellent agreement was found for 14 of 31 features (45.2%), with eight achieving excellent agreement (Gwet's AC >0.75) and seven of them being melanomaspecific features. These features were peppering/granularity (0.91), shiny white streaks (0.89), typical pigment network (0.83), blotch irregular (0.82), negative network (0.81), irregular globules (0.78), dotted vessels (0.77), and blue-whitish veil (0.76). By utilizing an exemplar dataset, a good-to-excellent agreement was found for 14 features that have previously been shown useful in discriminating nevi from melanoma. All images are public (www.isic-archive.com) and can be used for education, scientific communication, and machine learning experiments.
  • No Thumbnail Available
    Item
    Generation of a Melanoma and Nevus Data Set From Unstandardized Clinical Photographs on the Internet
    (2023) Cho, Soo Ick; Navarrete-Dechent, Cristian; Daneshjou, Roxana; Cho, Hye Soo; Chang, Sung Eun; Kim, Seong Hwan; Na, Jung-Im; Han, Seung Seog
    ImportanceArtificial intelligence (AI) training for diagnosing dermatologic images requires large amounts of clean data. Dermatologic images have different compositions, and many are inaccessible due to privacy concerns, which hinder the development of AI.ObjectiveTo build a training data set for discriminative and generative AI from unstandardized internet images of melanoma and nevus.Design, Setting, and ParticipantsIn this diagnostic study, a total of 5619 (CAN5600 data set) and 2006 (CAN2000 data set; a manually revised subset of CAN5600) cropped lesion images of either melanoma or nevus were semiautomatically annotated from approximately 500 000 photographs on the internet using convolutional neural networks (CNNs), region-based CNNs, and large mask inpainting. For unsupervised pretraining, 132 673 possible lesions (LESION130k data set) were also created with diversity by collecting images from 18 482 websites in approximately 80 countries. A total of 5000 synthetic images (GAN5000 data set) were generated using the generative adversarial network (StyleGAN2-ADA; training, CAN2000 data set; pretraining, LESION130k data set).Main Outcomes and MeasuresThe area under the receiver operating characteristic curve (AUROC) for determining malignant neoplasms was analyzed. In each test, 1 of the 7 preexisting public data sets (total of 2312 images; including Edinburgh, an SNU subset, Asan test, Waterloo, 7-point criteria evaluation, PAD-UFES-20, and MED-NODE) was used as the test data set. Subsequently, a comparative study was conducted between the performance of the EfficientNet Lite0 CNN on the proposed data set and that trained on the remaining 6 preexisting data sets.ResultsThe EfficientNet Lite0 CNN trained on the annotated or synthetic images achieved higher or equivalent mean (SD) AUROCs to the EfficientNet Lite0 trained using the pathologically confirmed public data sets, including CAN5600 (0.874 [0.042]; P = .02), CAN2000 (0.848 [0.027]; P = .08), and GAN5000 (0.838 [0.040]; P = .31 [Wilcoxon signed rank test]) and the preexisting data sets combined (0.809 [0.063]) by the benefits of increased size of the training data set.Conclusions and RelevanceThe synthetic data set in this diagnostic study was created using various AI technologies from internet images. A neural network trained on the created data set (CAN5600) performed better than the same network trained on preexisting data sets combined. Both the annotated (CAN5600 and LESION130k) and synthetic (GAN5000) data sets could be shared for AI training and consensus between physicians.
  • No Thumbnail Available
    Item
    Importance of Both Clinical and Dermoscopic Findings in Predicting High-Risk Histopathological Subtype in Facial Basal Cell Carcinomas
    (2024) Ceder, Hannah; Backman, Eva; Marghoob, Ashfaq; Navarrete-Dechent, Cristian; Polesie, Sam; Reiter, Ofer; Paoli, John
    Introduction: Being able to recognize high-risk facial basal cell carcinoma (BCC) may lead to fewer incomplete excisions and inappropriate treatments. Objectives: We sought to investigate clinical and dermoscopic criteria for predicting facial BCC subtypes, analyze the interobserver agreement between readers, and develop a diagnostic algorithm to predict high-risk histopathological subtype. Methods: In this single-center, retrospective investigation, 6 independent readers evaluated predefined clinical and dermoscopic criteria in images of histopathologically verified primary facial BCCs including: topography, border demarcation, vessels, ulceration, white porcelain areas, shiny white blotches and strands, and pigmented structures and vessels within ulceration. Results: Overall, 297 clinical and dermoscopic image pairs were analyzed. The strongest associations with high-risk subtype were: "bumpy" topography (OR 3.8, 95% CI, 3.1-4.7), ill-defined borders (OR 3.4, 95% CI 3.1-4.7), white porcelain area (OR 3.5, 95% CI 2.8-4.5), and vessels within ulceration (OR 3.1, 95% CI 2.4-4.1). Predominantly focused vessels were a positive diagnostic criterium for either nodular (OR 1.7, 95% CI 1.3-2.2) or high-risk (OR 2.0, 95% CI 1.6-2.5) subtypes and a strong negative diagnostic criterium for superficial BCC (OR 14.0, 95% CI 9.6-20.8). Interobserver agreement ranged from fair to substantial (kappa = 0.36 to 0.72). A diagnostic algorithm based on these findings demonstrated a sensitivity of 81.4% (95% CI, 78.9-83.7%) and a specificity of 53.3% (95% CI, 49.7-56.9%) for predicting high-risk BCC subtype. Conclusions: Integration of both clinical and dermoscopic features (including novel features such as topography and vessels within ulceration) are essential to improve subtype prediction of facial BCCs and management decisions.
  • No Thumbnail Available
    Item
    In vivo imaging characterization of basal cell carcinoma and cutaneous response to high-dose ionizing radiation therapy: A prospective study of reflectance confocal microscopy, dermoscopy, and ultrasonography
    (2021) Navarrete-Dechent, Cristian; Cordova, Miguel; Liopyris, Konstantinos; Aleissa, Saud; Rajadhyaksha, Milind; Cohen, Gil'ad; Marghoob, Ashfaq A.; Rossi, Anthony M.; Barker, Christopher A.
    Background: Radiation therapy (RT) is a treatment option for select skin cancers. The histologic effects of RT on normal skin or skin cancers are not well characterized. Dermoscopy, high-frequency ultrasonography (HFUS), and reflectance confocal microscopy (RCM) are noninvasive imaging modalities that may help characterize RT response.