Fabienne Fröhlich

Dr. Fabienne Fröhlich

orcid

Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
24 August 2023 in JMIR Dermatology
Verified
Journal Article
Open access
MGMathias Gassner
Mathias Gassner
JGJavier Barranco Garcia
Javier Barranco Garcia
STStephanie Tanadini-Lang
Stephanie Tanadini-Lang
FBFabio Bertoldo
Fabio Bertoldo
FFFabienne Fröhlich
Fabienne Fröhlich
RBRalph P Braun
Ralph P Braun
Background: Previous research studies have demonstrated that medical content image retrieval can play an important role by assisting dermatologists in skin lesion diagnosis. However, current state-of-the-art approaches have not been adopted in routine consultation, partly due to the lack of interpretability limiting trust by clinical users. Objective: This study developed a new image retrieval architecture for polarized or dermoscopic imaging guided by interpretable saliency maps. This approach provides better feature extraction, leading to better quantitative retrieval performance as well as providing interpretability for an eventual real-world implementation. Methods: Content-based image retrieval (CBIR) algorithms rely on the comparison of image features embedded by convolutional neural network (CNN) against a labeled data set. Saliency maps are computer vision–interpretable methods that highlight the most relevant regions for the prediction made by a neural network. By introducing a fine-tuning stage that includes saliency maps to guide feature extraction, the accuracy of image retrieval is optimized. We refer to this approach as saliency-enhanced CBIR (SE-CBIR). A reader study was designed at the University Hospital Zurich Dermatology Clinic to evaluate SE-CBIR’s retrieval accuracy as well as the impact of the participant’s confidence on the diagnosis. Results: SE-CBIR improved the retrieval accuracy by 7% (77% vs 84%) when doing single-lesion retrieval against traditional CBIR. The reader study showed an overall increase in classification accuracy of 22% (62% vs 84%) when the participant is provided with SE-CBIR retrieved images. In addition, the overall confidence in the lesion’s diagnosis increased by 24%. Finally, the use of SE-CBIR as a support tool helped the participants reduce the number of nonmelanoma lesions previously diagnosed as melanoma (overdiagnosis) by 53%. Conclusions: SE-CBIR presents better retrieval accuracy compared to traditional CBIR CNN-based approaches. Furthermore, we have shown how these support tools can help dermatologists and residents improve diagnosis accuracy and confidence. Additionally, by introducing interpretable methods, we should expect increased acceptance and use of these tools in routine consultation.

27 July 2023 in Cancers
Verified
Journal Article
Open access
PDF file
FFFabienne Fröhlich
Fabienne Fröhlich
EREgle Ramelyte
Egle Ramelyte
PTPatrick Turko
Patrick Turko
ADAndreas Dzung
Andreas Dzung
SFSandra N. Freiberger
Sandra N. Freiberger
Reinhard DummerReinhard Dummer
Reinhard Dummer
Novel treatment modalities comprising immune checkpoint inhibitors and targeted therapies have revolutionized treatment of metastatic melanoma. Still, some patients suffer from rapid progression and decease within months after a diagnosis of stage IV melanoma. We aimed to assess whether genomic alterations may predict survival after the development of stage IV disease, irrespective of received therapy. We analyzed tumor samples of 79 patients with stage IV melanoma using a custom next-generation gene-sequencing panel, MelArray, designed to detect alterations in 190 melanoma-relevant genes. We classified the patients: first, as short survivors (survival ≤6 months after stage IV disease, n = 22) and long survivors (survival >6 months, n = 57); second, by using a cut-off of one year; and third, by comparing the longest surviving 20 patients to the shortest surviving 20. Among analyzed genes, no individual gene alterations, or combinations of alterations, could be dichotomously associated with survival. However, the cohort’s mutational profiles closely matched three known mutational signatures curated by the Catalog of Somatic Mutations in Cancer (COSMIC): UV signature COSMIC_7 (cosine-similarity 0.932), clock-like signature COSMIC_5 (cosine-similarity 0.829), and COSMIC_30 (cosine-similarity 0.726). Patients with UV signature had longer survival compared to patients with clock-like and COSMIC 30 (p < 0.0001). Subgroup dichotomization at 6 months showed that 75% of patients with UV signature survived longer than 6 months, and about 75% of patients with clock-like signature survived less than 6 months after development of stage IV disease. In our cohort, clock-like COSMIC_5 mutational signature predicted poor survival while a UV signature COSMIC_7 predicted longer survival. The prognostic value of mutational signatures should be evaluated in prospective studies.

19 November 2025 in Biomedicines
Verified
Journal Article
Open access
PDF file
PPPatrick Poffet
Patrick Poffet
VBVeronika Baghin
Veronika Baghin
FFFabienne Fröhlich
Fabienne Fröhlich
IBIrina Banzola
Irina Banzola
JLJulia Laube
Julia Laube
BMBarbara Meier-Schiesser
Barbara Meier-Schiesser
Background/Objectives: Genital lichen sclerosus (LS) and lichen planus genitalis (Lpg) are chronic inflammatory dermatoses with overlapping clinical features but incompletely understood pathogenesis. Current therapies are largely symptomatic. Methods: To clarify underlying mechanisms and identify therapeutic targets, we retrospectively analyzed 174 patients (142 LS and 32 Lpg). Clinical features and comorbidities were compared, and gene expression profiling of 730 inflammation-related genes was performed on lesional tissue from LS and Lpg patients and healthy controls using NanoString technology. Selected findings were validated by immunohistochemistry. Results: LS patients were older and predominantly female and more frequently had metabolic syndrome. On the molecular level, LS showed a generally lower inflammatory gene expression profile than Lpg. Nevertheless, LS was characterized by strong upregulation of CCL27 and MARCO, whereas Lpg displayed enhanced IL-1 pathway activation and increased expression of B-cell–associated markers. Conclusions: These results demonstrate distinct immunological differences between the two conditions and provide further insight into disease-specific pathways.

24 August 2022 in JMIR Preprints
Verified
Preprint
Open access
MGMathias Gassner
Mathias Gassner
JGJavier Barranco Garcia
Javier Barranco Garcia
STStephanie Tanadini-Lang
Stephanie Tanadini-Lang
FBFabio Bertoldo
Fabio Bertoldo
FFFabienne Fröhlich
Fabienne Fröhlich
RBRalph P Braun
Ralph P Braun
BACKGROUND Previous research studies have demonstrated that medical content image retrieval can play an important role by assisting dermatologists in skin lesion diagnosis. However, current state-of-the-art approaches have not been adopted in routine consultation, partly due to the lack of interpretability limiting trust by clinical users. OBJECTIVE This study developed a new image retrieval architecture for polarized or dermoscopic imaging guided by interpretable saliency maps. This approach provides better feature extraction, leading to better quantitative retrieval performance as well as providing interpretability for an eventual real-world implementation. METHODS Content-based image retrieval (CBIR) algorithms rely on the comparison of image features embedded by convolutional neural network (CNN) against a labeled data set. Saliency maps are computer vision–interpretable methods that highlight the most relevant regions for the prediction made by a neural network. By introducing a fine-tuning stage that includes saliency maps to guide feature extraction, the accuracy of image retrieval is optimized. We refer to this approach as saliency-enhanced CBIR (SE-CBIR). A reader study was designed at the University Hospital Zurich Dermatology Clinic to evaluate SE-CBIR’s retrieval accuracy as well as the impact of the participant’s confidence on the diagnosis. RESULTS SE-CBIR improved the retrieval accuracy by 7% (77% vs 84%) when doing single-lesion retrieval against traditional CBIR. The reader study showed an overall increase in classification accuracy of 22% (62% vs 84%) when the participant is provided with SE-CBIR retrieved images. In addition, the overall confidence in the lesion’s diagnosis increased by 24%. Finally, the use of SE-CBIR as a support tool helped the participants reduce the number of nonmelanoma lesions previously diagnosed as melanoma (overdiagnosis) by 53%. CONCLUSIONS SE-CBIR presents better retrieval accuracy compared to traditional CBIR CNN-based approaches. Furthermore, we have shown how these support tools can help dermatologists and residents improve diagnosis accuracy and confidence. Additionally, by introducing interpretable methods, we should expect increased acceptance and use of these tools in routine consultation.

Verified
Journal Article
Open access
PDF file
KGKen Suemer Goekcimen
Ken Suemer Goekcimen
CKCadri Knoch
Cadri Knoch
FFFabienne Fröhlich
Fabienne Fröhlich
TKThomas Martin Kündig
Thomas Martin Kündig
CGChristian Greis
Christian Greis
BMBarbara Meier-Schiesser
Barbara Meier-Schiesser
Background: Lichen planus (LP) is a chronic inflammatory dermatosis with multiple clinical variants involving the skin, mucous membranes, nails, and hair follicles. Methods: We conducted a retrospective, descriptive study of patients diagnosed with LP at a tertiary referral center from January 2011 to December 2024. Inclusion required concordance between clinical presentation and histopathologic findings. Demographic characteristics, LP subtypes, anatomical involvement, comorbidities, therapeutic approaches, and treatment outcomes were extracted from electronic health records. In addition, an exploratory sensitivity analysis restricted to patients with a single LP subtype was performed to allow for independent subgroup comparisons, and a modified Charlson Comorbidity Index (CCI) based on available comorbidity domains was calculated. Pairwise multivariable logistic regression models adjusted for age, sex, and outcome-specific modified CCI were performed for selected comorbidities. Results: A total of 754 patients were included (mean age 53.1 years), with cutaneous LP (cLP), oral LP (oLP), genital LP (gLP), and lichen planopilaris (LPP) being the most frequent subtypes; a total of 620 had a single major LP subtype and were included in the mutually exclusive analysis. In these groups, modified CCI, age-adjusted modified CCI, and overall comorbidity count differed significantly across subtypes (Kruskal–Wallis p < 0.001). After adjustment in pairwise models, cLP-only showed higher odds of malignancy compared with oLP-only, gLP-only, and LPP-only and higher odds of diabetes mellitus compared with all other pure subtypes. Most other comorbidity comparisons were non-significant or imprecise because of low event numbers. Topical glucocorticoids were the most frequently used treatment, and treatment responses varied by subtype, being more effective in cLP and gLP compared to LPP. Topical calcineurin inhibitors demonstrated the highest response rates in gLP. Acitretin was most effective in cLP, whereas isotretinoin showed favorable responses in oLP. Conclusions: This large, histopathologically confirmed cohort highlights distinct differences in comorbidity patterns, anatomical involvement, and therapeutic response across LP subtypes. Treatment outcomes vary substantially by subtype, underscoring the need for individualized management strategies. Prospective studies are warranted to further elucidate subtype-specific disease associations and optimize treatment approaches.

19 October 2022
Verified
Conference Paper
Closed access
FFFabienne Fröhlich
Fabienne Fröhlich
Background: Pyoderma gangrenosum (PG) belongs to the group of neutrophilic dermatoses and is characterized by painful ulcerations. The underlying |