Liutkus, Jokūbas

Mobiliąsias programas nešiojamuosiuose įrenginiuose naudoti daug lengviau tiek pacientams, tiek gydytojams. Tačiau tik keliuose tyrimuose naudoti išmanieji telefonai ir juose integruotas dirbtinio intelekto (DI) algoritmas vienos kategorijos piktybinio odos naviko klasifikacijai. Vykdant DIVAT-SMARTPHONE projektą sukurta odos vaizdų siuntimo programa išoriniam vartotojui, suintegruota su LSMU ligoninės Kauno klinikų Pacientų portalu. Sumani technologija, akredituota kaip A klasės medicinos priemonė, leidžia teikti pacientams teledermatologijos paslaugas ir kaupti odos ligų vaizdus dirbtinio intelekto (DI) programos validavimui ir tobulinimui. Dirbtinio intelekto analizės technologija (DIVAT) sukurta konvoliucinio neuroninio tinklo You Only Look Once pagrindu. Panaudoti vaizdai iš išorinių (n=58457) ir vidinės (n=633) duomenų bazių. Dirbtinio intelekto algoritmas patikrintas analizuojant odos ligų vaizdus (n=100), kurių klinikinė diagnozė patvirtinta remiantis histologiniu tyrimu. DIVAT technologija pagerins konsultacinės pagalbos prieinamumą pas medicinos specialistą, nes ją galima taikyti rutininėje praktikoje teikiant rekomendacinio pobūdžio išvadą dėl odos ligos diagnozės.

Introduction & Objectives: Our innovative artificial intelligence algorithm (DIVAT), based on a “You Only Look Once” (YOLO) neural network model, is intended to enhance the accuracy of store-and-forward teledermatology services. We present the effectiveness of DIVAT in the classification of melanoma, melanocytic nevi, and seborrheic keratoses, comparing the results with two dermoscopy experts and five beginners. The research is funded by the Research Council of Lithuania COVID-19 R&D Implementation scheme. Materials & Methods: To train the DIVAT algorithm, we utilized a robust dataset consisting of dermoscopic images from the International Skin Imaging Collaboration (ISIC) archive, which contained 58,457 images. This dataset was augmented with our department’s own dermoscopic images, which consisted of 633 images, 251 of which were histopathologically verified. Following training, we integrated the DIVAT algorithm into a mobile application, which was used for subsequent assessment. The testing process involved analyzing 100 dermoscopic images of histologically confirmed melanomas (n=32), melanocytic nevi (n=35), and seborrheic keratoses (n=33). To benchmark the performance of DIVAT, we compared its results to the gold standard of histologically confirmed diagnosis, as well as evaluations by two skilled dermatologists with extensive experience in dermoscopy and five dermoscopy beginners. Results: DIVAT achieved the highest sensitivity when classifying melanoma, with a sensitivity of 0.88 (0.71 – 0.96) and a specificity of 0.87 (0.76 – 0.94). The algorithm also performed well in classifying melanocytic nevi, with a sensitivity of 0.77 (0.60 – 0.90) and a specificity of 0.91 (0.81-0.97). However, the NNM sensitivity for seborrheic keratoses was lower, at 0.52 (0.34 – 0.69), although it maintained a high specificity of 0.93 (0.83-0.98) (Table 1). [...]. Conclusion: DIVAT’s classification performance for melanoma and melanocytic nevi was superior to the beginner raters. However, its performance in classifying seborrheic keratoses was lower than that of both the beginner and skilled raters, which we attribute to the smaller amount of labeled training data available for this lesion type.

Current artificial intelligence algorithms can classify melanomas at a level equivalent to that of experienced dermatologists. The objective of this study was to assess the accuracy of a smartphone-based "You Only Look Once" neural network model for the classification of melanomas, melanocytic nevi, and seborrheic keratoses. The algorithm was trained using 59,090 dermatoscopic images. Testing was performed on histologically confirmed lesions: 32 melanomas, 35 melanocytic nevi, and 33 seborrheic keratoses. The results of the algorithm's decisions were compared with those of two skilled dermatologists and five beginners in dermatoscopy. The algorithm's sensitivity and specificity for melanomas were 0.88 (0.71-0.96) and 0.87 (0.76-0.94), respectively. The algorithm surpassed the beginner dermatologists, who achieved a sensitivity of 0.83 (0.77-0.87). For melanocytic nevi, the algorithm outclassed each group of dermatologists, attaining a sensitivity of 0.77 (0.60-0.90). The algorithm's sensitivity for seborrheic keratoses was 0.52 (0.34-0.69). The smartphone-based "You Only Look Once" neural network model achieved a high sensitivity and specificity in the classification of melanomas and melanocytic nevi with an accuracy similar to that of skilled dermatologists. However, a bigger dataset is required in order to increase the algorithm's sensitivity for seborrheic keratoses.

Quarantines and associated lockdowns during the COVID-19 pandemic have significantly increased the Breslowthickness in primary cutaneous melanomas (CM) due to curbed patient access to medical specialists [1], highlightingthe importance of timely diagnostics. Teledermatology can be used to expedite this process and has been shown toreduce patients’ waiting times for specialist consultations and lower the associated health care costs [2]. Routinely,primary CM is diagnosed on the basis of presentation and dermatoscopy, but promising results for the differentiationof melanocytic skin tumours were also observed using spectrophotometric (SIAscope) features, which correspond tothe distribution, location and concentration of skin chromophores [3]. CM thickness, measured after tumour excisionaccording to Breslow (pT), is an important prognostic indicator for sentinel lymph node biopsy. Previous studies haveshown that CM thickness before surgery can be accurately estimated using 14 MHz ultrasound when pT is > 2 mm[4], and similar sensitivities can be achieved for CM pT of less than 2 mm with a 22 MHz ultrasound probe [5,6]. Skinimaging technologies are important aids for CM diagnostics and their incorporation into computer algorithms forclinical decision support using artificial intelligence can further speed up and facilitate CM diagnostics before surgery.The availability of smartphone-attachable dermoscopy devices continues to grow and the advanced capabilities ofmodern smartphone allow these technologies to be implemented into freely available handheld systems. As a result,there is an increased need for mobile applications that would allow artificial intelligence guided analysis ofsmartphone-captured images at the time of patient care [7].

Introduction Hand-foot syndrome (HFS), also known as palmoplantar erythrodysesthesia or acral erythema, is a well-known adverse reaction to systemic chemotherapy. Clinical presentation includes palmoplantar pain, dysesthesia, or numbness, and is frequently followed by demarcated erythema, edema, fissuring, or scaling. High-grade events may include blistering and ulceration. The most common culprit medications include liposomal doxorubicin, docetaxel, 5-fluorouracil, and its prodrug capecitabine. Case report An 81-year-old male presented to the dermatology department due to severe pain and lesions in soles of both feet. From documentation it was discovered that the patient was diagnosed with stage III sigmoid colon adenocarcinoma and had received a 7th course of capecitabine chemotherapy two days prior. Upon clinical examination, sharply demarcated erythema and complete plantar desquamation was seen. A grade 4 capecitabine-induced hand-foot syndrome was diagnosed. The patient was started on a 30mg prednisone taper and wound treatment with topical chlorhexidine irrigation and paraffin gauze dressing applications every two days. Upon improvement, the patient was discharged and outpatient treatment with hydrocolloid dressings was recommended. The case was discussed in a multidisciplinary tumor board and further adjuvant treatment was discontinued. There were no HFS exacerbations after withdrawal of capecitabine, with a continuous reduction in HFS severity and improvement in the patient’s Activities of Daily Living (ADL). Conclusion HFS is a common side effect of capecitabine and can rarely present as a high-grade adverse event, requiring prompt treatment and modification of systemic chemotherapy regimens. Anticipation and early diagnostics can ensure a reduction in morbidity and an improvement in patients’ quality of life.

Introduction Bullous pemphigoid (BP) is an autoimmune subepidermal blistering disorder that most commonly occurs in the elderly. Some authors have described several cases of BP associated with malignancies, but the evidence of this correlation remains controversial. A possible association between malignant neoplasms and BP is the formation of tumor-directed antibodies, which cross-react with antigens in the basement membrane zone leading to the formation of blisters [1]. Aims and Objectives The aim is to present a rare case of locally advanced basal cell carcinoma (BCC) in an elderly patient with BP. Materials and methods An 84-year-old female patient was admitted to the hospital because of itchy, widespread bullae, fever (38.9 °C), and a painful 25 x 10 cm sized ulcer under both breasts. During the dermatological examination, we found tense bullae with a negative Nikolsky sign. There were no lesions in the mucosa. The patients’ history revealed that the ulcer, which had never been treated before, appeared 15 years ago and the bullae began to form within the past 3 weeks. In addition, the patient’s health condition was aggravated by chronic autoimmune hemolytic anemia which was systemically treated with prednisolone and splenectomy in the hematology department 15 years ago. Results A punch biopsy of the bullae and histopathological analysis using hematoxylin and eosin staining showed a subepidermal blister and dense perivascular infiltrate. Direct immunofluorescence of perilesional skin displayed linear deposition of IgG and C3 along the basement membrane, consistent with BP. Histopathological analysis of the ulcer revealed multifocal nests of atypical basaloid epithelium with peripheral palisading and cleft formation, supporting the diagnosis of BCC. Additionally, cultures from the ulcer and bullae grew Staphylococcus aureus. The patient was treated with intravenous methylprednisolone 500 mg daily for 4 days, followed by oral prednisolone taper over 13

Introduction Kaposi sarcoma (KS) is an endothelial tumor associated with human herpes virus 8 (HHV-8) infection. The iatrogenic form of KS arises due to immunosuppressive treatment and is most common following solid organ transplantation. Patients typically present with angiomatoid lesions on the lower limbs and lymphedema. Currently, there are no guidelines specific for iatrogenic KS diagnostics and treatment, although a change in the immunosuppressive regimen is often employed and is associated with lesion resolution. Aims and Objectives We present a case of iatrogenic KS following a renal allograft, with rapid progression mandating systemic treatment. Materials and methods A 74-year-old female was referred to the dermatology department due to expanding itchy tumors on the right leg with a duration of 5 months. Previous treatment with local corticosteroids and H1 antihistamines was ineffective. The patient had received a cadaveric renal allograft two years before and the immunosuppressive therapy consisted of cyclosporine 175mg and methylprednisolone 4mg q.d. She was HIV negative with a normal CD4 cell count. On dermatological examination, multiple grouped red-violet nodules on the right thigh and lower leg were seen. There was associated unilateral limb edema. Peripheral lymph nodes were not palpable. On dermoscopy, redviolet structureless homogenous zones with KS-specific rainbow pattern were identified. Results A lesion biopsy displayed a dermal tumor composed of spindle-shaped CD31+ CD34+ cells with positive staining for HHV-8 and slit-like spaces containing erythrocytes, consistent with KS nodular stage.Computer tomography scans of the chest, abdomen and pelvis showed peritoneal carcinomatosis with multiple affected mesenteric lymph nodes up to 1.2 cm, in addition to the involved lower thoracic paravertebral, para-aortic and hepatic hilar lymph nodes up to 1.1 cm in size. Following negative cytomegalovirus DNA serology, an ultrasound guided mesenteric

Introduction Cutaneous melanoma (CM) is the ninth most common type of cancer and the second leading cause of death. Localized CM has a 99% 5-year survival rate, but in patients with distant metastases the 5-year survival rate decreases to 20%. The incidence and mortality of CM in Lithuania has been increasing for the last three decades due to the lack of early diagnosis [1] and delay in reimbursement of targeted therapy with BRAFi + MEKi inhibitors and anti- PD1 immunotherapy for progressive stages of disease [2]. Aims and Objectives To review diagnostics and monitoring of CM based on own experience and recent scientific literature. Materials and methods Results of our clinical studies were analysed and articles containing terms for diagnostics and monitoring of CM were retrieved to generate an up-todate view on this topic. Results Routinely, primary CM is diagnosed based on clinical and dermatoscopic features, but promising results were observed using spectrophotometry (SIAscopy), which shows the distribution of skin chromophores (including melanin, hemoglobin, and collagen) in the epidermis and dermis [3]. CM thickness, measured histopathologically (pT) after tumour excision based on the 8th version of AJCC classification, is an important prognostic indicator (in cases of pT >=0.8 mm) for sentinel lymph node (SLN) biopsy [1]. Our studies have shown that CM thickness before excisional biopsy can be accurately estimated using 14 MHz ultrasound when pT is >2 mm [4], and similar accuracy can be achieved for pT of less than 2 mm with a 22 MHz ultrasound probe [5,6]. CM up to 0.8 mm in pT does not need further imaging diagnostics (ID) for staging. From stage IB onwards, patient examination with lymph node sonography (LNS) is recommended, with no further ID. From stage IIC whole-body computed tomography (CT) and/or positron emission tomography (PET)-CT in combination with brain MRI should be performed. From stage III and higher, molecular testing for BRAF V600 is

Introduction. A case report of a patient diagnosed with angiomatoid fibrous histiocytoma, with rainbow pattern sign on dermoscopy. Case Description. A 31-year-old HIV-negative patient was referred to a tertiary dermatology clinic due to a slow-growing lesion on the right forearm, which bled upon trauma, present for 2 years. Physical examination revealed a blue-grey nodule 11x8 mm in size. On dermoscopy, a central brown structureless area with white structures, peripheral bluegray homogenous area, regular radial streaming, rainbow pattern and crystalline structures were seen. Siascopic examination registered blood displacement with erythematous blush and dermal melanin. Wide local excision and histopathologic investigation were performed. Microscopically, a poorly demarcated tumor of the reticular dermis was seen, with nests of spindle and ovoid cells dispersed within the collagen fibers, Touton giant cells, sparse erythrocyte-filled cystic spaces and abundant hemosiderophages. Diagnosis of angiomatoid fibrous histiocytoma was established, with no recurrence at 6 months follow up. Summary. The rainbow pattern dermoscopic sign present in angiomatoid fibrous histiocytoma – a rare soft tissue neoplasm with histologic features of nodular distribution of ovoid and spindle cells with blood-filled cystic cavities. Few specific clinical and dermoscopic features have been previously described, making early diagnosis challenging. Conclusion. Angiomatoid fibrous histiocytoma (AFH) is a rare tumor of intermediate biological behavior, most commonly occurring in children and young adults. It is difficult to diagnose clinically, with no established dermoscopic criteria. To our knowledge, this is the first description of the rainbow pattern dermoscopic sign for AFH, a feature most strongly associated [...]

The incidence and mortality of cutaneous melanoma (CM) in Lithuania has been increasing for the last three decades due to the lack of early diagnosis [1] and delays in reimbursement of targeted therapy with BRAFi + MEKi inhibitors and anti-PD1 immunotherapy for progressive stages of disease [2]. Routinely, primary CM is diagnosed on the basis of presentation and dermatoscopy, but promising results for differentiation of melanocytic skin tumours were also observed using spectrophotometric (SIAscope) features, which correspond to the distribution, location and concentration of skin chromophores [3]. CM thickness, measured after tumour excision according to Breslow (pT), is an important prognostic indicator for sentinel lymph node biopsy. Previous studies have shown that CM thickness before surgery can be accurately estimated using 14 MHz ultrasound when pT is > 2 mm [4], and similar sensitivities can be achieved for CM pT of less than 2 mm with a 22 MHz ultrasound probe [5,6]. Such non-invasive skin imaging technologies are important aids for CM diagnostics and their incorporation into computer algorithms for clinical decision support based on artificial intelligence can further speed up and facilitate CM diagnostics before surgery.