Hülya Efetürk
Research Subjects
Hülya Efetürk has a total of 31 co-authors. Here are the top co-authors by number of shared publications:
- Aleksander Kosmala
Department of Nuclear Medicine, University Hospital Würzburg, 97080 Würzburg, Germany
2 shared publications - Andreas Konrad Buck
Department of Nuclear Medicine, University Hospital Wuerzburg, Wuerzburg, Germany
2 shared publications - Aykut Kürşat Fidan
Samsun Gazi Devlet Hastanesi
2 shared publications - Bedriye Büşra Demirel
Department of Nuclear Medicine, Ankara Oncology Research and Training Hospital, Ankara, Turkey
2 shared publications - Gülin Uçmak
SAĞLIK BİLİMLERİ ÜNİVERSİTESİ, ANKARA DR. ABDURRAHMAN YURTASLAN ONKOLOJİ SAĞLIK UYGULAMA VE ARAŞTIRMA MERKEZİ
2 shared publications - İpEk Öztürk
Department of Nuclear Medicine, Başakşehir Cam and Sakura City Hospital, Istanbul, Turkey
2 shared publications
Latest Publications
- Open Access
Detection of Local Prostate Cancer Recurrence from PET/CT Scans Using Deep Learning
Background: Prostate cancer (PC) is a leading cause of cancer-related deaths in men worldwide. PSMA-directed positron emission tomography (PET) has shown promising results in detecting recurrent PC and metastasis, improving the accuracy of diagnosis and treatment planning. To evaluate an artificial intelligence (AI) model based on [18F]-prostate specific membrane antigen (PSMA)-1007 PET datasets for the detection of local recurrence in patients with prostate cancer. Methods: We retrospectively analyzed 1404 [18F]-PSMA-1007 PET/CTs from patients with histologically confirmed prostate cancer. Artificial neural networks were trained to recognize the presence of local recurrence based on the PET data. First, the hyperparameters were optimized for an initial model (model A). Subsequently, the bladder was localized using an already published model and a model (model B) was trained only on a 20 cm cube around the bladder. Finally, two separate models were trained on the same section depending on the prostatectomy status (model C (post-prostatectomy) and model D (non-operated)). Results: Model A achieved an accuracy of 56% on the validation data. By restricting the region to the area around the bladder, Model B achieved a validation accuracy of 71%. When validating the specialized models according to prostatectomy status, model C achieved an accuracy of 77% and model D an accuracy of 77%. All models achieved accuracies of almost 100% on the training data, indicating overfitting. Conclusions: For the presented task, 1404 examinations were insufficient to reach an accuracy of over 90% even when employing data augmentation, including additional metadata and performing automated hyperparameter optimization. The low F1-score and AUC values indicate that none of the presented models produce reliable results. However, we will facilitate future research and the development of better models by openly sharing our source code and all pre-trained models for transfer learning.
Open AccessDetection of Local Prostate Cancer Recurrence from PET/CT Scans Using Deep Learning
Background: Prostate cancer (PC) is a leading cause of cancer-related deaths in men worldwide. PSMA-directed positron emission tomography (PET) has shown promising results in detecting recurrent PC and metastasis, improving the accuracy of diagnosis and treatment planning. To evaluate an artificial intelligence (AI) model based on [18F]-prostate specific membrane antigen (PSMA)-1007 PET datasets for the detection of local recurrence in patients with prostate cancer. Methods: We retrospectively analyzed 1404 [18F]-PSMA-1007 PET/CTs from patients with histologically confirmed prostate cancer. Artificial neural networks were trained to recognize the presence of local recurrence based on the PET data. First, the hyperparameters were optimized for an initial model (model A). Subsequently, the bladder was localized using an already published model and a model (model B) was trained only on a 20 cm cube around the bladder. Finally, two separate models were trained on the same section depending on the prostatectomy status (model C (post-prostatectomy) and model D (non-operated)). Results: Model A achieved an accuracy of 56% on the validation data. By restricting the region to the area around the bladder, Model B achieved a validation accuracy of 71%. When validating the specialized models according to prostatectomy status, model C achieved an accuracy of 77% and model D an accuracy of 77%. All models achieved accuracies of almost 100% on the training data, indicating overfitting. Conclusions: For the presented task, 1404 examinations were insufficient to reach an accuracy of over 90% even when employing data augmentation, including additional metadata and performing automated hyperparameter optimization. The low F1-score and AUC values indicate that none of the presented models produce reliable results. However, we will facilitate future research and the development of better models by openly sharing our source code and all pre-trained models for transfer learning.
- Open Access
- 3 March 2025
- Published by MDPI AG in Preprints.org
Detection of Local Prostate Cancer Recurrence from PET/CT Scans Using Deep Learning
Prostate cancer (PC) is a leading cause of cancer-related deaths in men worldwide. PSMA-directed positron emission tomography (PET) has shown promising results in detecting recurrent PC and metastasis. We developed a deep-learning model to predict cancer recurrence from PSMA PET/CT images of prostate cancer patients after treatment. Different methods were used to improve the performance of the initial model, such as modifying the region of interest (ROI), including metadata as additional layers, or passing prostatectomy state to the model. A hyperparameter optimization of multiple parameters was performed to further increase the model's performance, which, combined with including metadata as additional image layers, resulted in the best validation accuracy of 77 %. Even though the final validation accuracy fell short of reaching 90 %, significant improvements were made to the model, and many approaches were tested that can be a basis for further research and improvements toward the performance of a model that can confidently detect cancer remission in the prostate or prostate bed.
- 3 March 2025
- Published by MDPI AG in Preprints.org
Open AccessDetection of Local Prostate Cancer Recurrence from PET/CT Scans Using Deep Learning
Prostate cancer (PC) is a leading cause of cancer-related deaths in men worldwide. PSMA-directed positron emission tomography (PET) has shown promising results in detecting recurrent PC and metastasis. We developed a deep-learning model to predict cancer recurrence from PSMA PET/CT images of prostate cancer patients after treatment. Different methods were used to improve the performance of the initial model, such as modifying the region of interest (ROI), including metadata as additional layers, or passing prostatectomy state to the model. A hyperparameter optimization of multiple parameters was performed to further increase the model's performance, which, combined with including metadata as additional image layers, resulted in the best validation accuracy of 77 %. Even though the final validation accuracy fell short of reaching 90 %, significant improvements were made to the model, and many approaches were tested that can be a basis for further research and improvements toward the performance of a model that can confidently detect cancer remission in the prostate or prostate bed.
- Open Access
- 25 February 2025
- Published by Hitit University in Hitit Medical Journal
- Vol. 7 (1) , 45-52
The Diagnostic Accuracy of V/P Scintigraphy in Pulmonary Embolism and Superiority of V/P SPECT to V/P Planar Scintigraphy
Objective: The aim of this retrospective study is to investigate the diagnostic accuracy of planar V/P scintigraphy and V/P single-photon emission computed tomography (SPECT) in patients who referred to our clinic for V/P scintigraphy with prediagnosis of pulmonary embolism (PE), as well as to investigate the contribution of V/P SPECT technique to planar V/P technique. Material and Method: The records of 204 patients, who were preliminarily diagnosed with PE within 1 year, were retrospectively reviewed. In our investigation of the diagnostic accuracy of V/P scintigraphy in for PE, we excluded three patients who only underwent perfusion scintigraphy and 20 patients whose final diagnoses could not be confirmed. This left a total of 181 patients included in the statistical analysis. Furthermore, we evaluated the contribution of SPECT to planar imaging in 48 patients, for whom V/P Scintigraphy were reported as positive and whose final diagnoses confirmed PE. Results: The sensitivity, specificity, negative predictive value, positive predictive value, and accuracy rate of V/P SPECT were calculated as 98%, 94.7%, 99.2%, 87.3%, and 95.6%, respectively. For planar scintigraphy, they were found to be 71.4%, 95.4%, 90%, 85.3%, and 88.9%, respectively. In SPECT, 13 (27.1%) patients who were not compatible with pulmonary embolism (PE) on planar imaging were found to have findings consistent with PE. In nine patients (18.8%), additional defects not observed on planar imaging were identified. Although the goodness of fit with the final diagnosis of both methods was statistically significant, SPECT (95.6%) performed better than planar (88.9%) imaging. Conclusion: Consistent with previous studies, it was found that while both imaging methods were successful, SPECT demonstrated higher diagnostic accuracy than planar scintigraphy in diagnosing PE. Therefore, it can be hypothesized that V/P scintigraphy can be safely deemed the first-choice in the diagnosis of PE.
- 25 February 2025
- Published by Hitit University in Hitit Medical Journal
- Vol. 7 (1) , 45-52
Open AccessThe Diagnostic Accuracy of V/P Scintigraphy in Pulmonary Embolism and Superiority of V/P SPECT to V/P Planar Scintigraphy
Objective: The aim of this retrospective study is to investigate the diagnostic accuracy of planar V/P scintigraphy and V/P single-photon emission computed tomography (SPECT) in patients who referred to our clinic for V/P scintigraphy with prediagnosis of pulmonary embolism (PE), as well as to investigate the contribution of V/P SPECT technique to planar V/P technique. Material and Method: The records of 204 patients, who were preliminarily diagnosed with PE within 1 year, were retrospectively reviewed. In our investigation of the diagnostic accuracy of V/P scintigraphy in for PE, we excluded three patients who only underwent perfusion scintigraphy and 20 patients whose final diagnoses could not be confirmed. This left a total of 181 patients included in the statistical analysis. Furthermore, we evaluated the contribution of SPECT to planar imaging in 48 patients, for whom V/P Scintigraphy were reported as positive and whose final diagnoses confirmed PE. Results: The sensitivity, specificity, negative predictive value, positive predictive value, and accuracy rate of V/P SPECT were calculated as 98%, 94.7%, 99.2%, 87.3%, and 95.6%, respectively. For planar scintigraphy, they were found to be 71.4%, 95.4%, 90%, 85.3%, and 88.9%, respectively. In SPECT, 13 (27.1%) patients who were not compatible with pulmonary embolism (PE) on planar imaging were found to have findings consistent with PE. In nine patients (18.8%), additional defects not observed on planar imaging were identified. Although the goodness of fit with the final diagnosis of both methods was statistically significant, SPECT (95.6%) performed better than planar (88.9%) imaging. Conclusion: Consistent with previous studies, it was found that while both imaging methods were successful, SPECT demonstrated higher diagnostic accuracy than planar scintigraphy in diagnosing PE. Therefore, it can be hypothesized that V/P scintigraphy can be safely deemed the first-choice in the diagnosis of PE.
- Open Access
- 28 June 2021
- Published by The Scientific and Technological Research Council of Turkey (TUBITAK-ULAKBIM) - DIGITAL COMMONS JOURNALS in TURKISH JOURNAL OF MEDICAL SCIENCES
- Vol. 51 (3) , 1115-1122
The relation between staging fluorine-18 fluorodeoxyglucose positron emission tomog- raphy/computed tomography metabolic parameters and tumor necrosis rate in pediatric osteosarcoma patients
The aim of this study was to investigate the contribution of fluorine-18 (F-18) fluorodeoxyglucose (FDG) positron emission tomography/computed tomography (PET/CT) imaging in staging of pediatric osteosarcoma patients and also to evaluate the ability of metabolic parameters from the primary tumor to predict tumor necrosis rate (TNR). F-18 FDG-PET/CT imaging was performed in staging 37 pediatric osteosarcoma patients. The metabolic parameters SUVmax (maximum standardised uptake value), MTV (metabolic tumour volume) and TLG (total lesion glycolysis) were measured from the primary tumor. TNR level of the primary tumor was histopathologically measured after standard neoadjuvant chemotherapy treatment. The contribution of F-18 FDG-PET/CT to staging of pediatric osteosarcoma patients and the accuracy of metabolic parameters of the primary tumor to predict TNR were analized by regression analysis. MTV and TLG of the primary tumor were found to efficiently predict histopathologic TNR, whereas SUVmax was not (p=0.012, p=0.027, p=0.25, respectively). Also 5 of 12 patients (41.6%) who were initially defined as localised osteosarcoma were upstaged in consequence of staging F-18 FDG-PET/CT findings. F-18 FDG-PET/CT staging in pediatric osteosarcoma patients can effectively distinguish metastatic-localised disease. MTV and TLG values are important parameters which can efficiently be used to predict TNR.
- 28 June 2021
- Published by The Scientific and Technological Research Council of Turkey (TUBITAK-ULAKBIM) - DIGITAL COMMONS JOURNALS in TURKISH JOURNAL OF MEDICAL SCIENCES
- Vol. 51 (3) , 1115-1122
Open AccessThe relation between staging fluorine-18 fluorodeoxyglucose positron emission tomog- raphy/computed tomography metabolic parameters and tumor necrosis rate in pediatric osteosarcoma patients
The aim of this study was to investigate the contribution of fluorine-18 (F-18) fluorodeoxyglucose (FDG) positron emission tomography/computed tomography (PET/CT) imaging in staging of pediatric osteosarcoma patients and also to evaluate the ability of metabolic parameters from the primary tumor to predict tumor necrosis rate (TNR). F-18 FDG-PET/CT imaging was performed in staging 37 pediatric osteosarcoma patients. The metabolic parameters SUVmax (maximum standardised uptake value), MTV (metabolic tumour volume) and TLG (total lesion glycolysis) were measured from the primary tumor. TNR level of the primary tumor was histopathologically measured after standard neoadjuvant chemotherapy treatment. The contribution of F-18 FDG-PET/CT to staging of pediatric osteosarcoma patients and the accuracy of metabolic parameters of the primary tumor to predict TNR were analized by regression analysis. MTV and TLG of the primary tumor were found to efficiently predict histopathologic TNR, whereas SUVmax was not (p=0.012, p=0.027, p=0.25, respectively). Also 5 of 12 patients (41.6%) who were initially defined as localised osteosarcoma were upstaged in consequence of staging F-18 FDG-PET/CT findings. F-18 FDG-PET/CT staging in pediatric osteosarcoma patients can effectively distinguish metastatic-localised disease. MTV and TLG values are important parameters which can efficiently be used to predict TNR.
- Open Access
- 3 April 2018
- Published by Springer Nature in Alzheimer's Research & Therapy
- Vol. 10 (1) , 35
Metabolic correlates of reserve and resilience in MCI due to Alzheimer's Disease (AD)
We explored the presence of both reserve and resilience in late-converter mild cognitive impairment due to Alzheimer’s disease (MCI-AD) and in patients with slowly progressing amyloid-positive MCI by assessing the topography and extent of neurodegeneration with respect to both “aggressive” and typically progressing phenotypes and in the whole group of patients with MCI, grounding the stratification on education level. We analyzed 94 patients with MCI-AD followed until conversion to dementia and 39 patients with MCI who had brain amyloidosis (AMY+ MCI), all with available baseline 18F-fluorodeoxyglucose positron emission tomography (FDG-PET) results. Using a data-driven approach based on conversion time, patients with MCI-AD were divided into typical AD and late-converter subgroups. Similarly, on the basis of annual rate of Mini Mental State Examination score reduction, AMY+ MCI group was divided, obtaining smoldering (first tertile) and aggressive (third tertile) subgroups. Finally, we divided the whole group (MCI-AD and AMY+ MCI) according to years of schooling, obtaining four subgroups: poorly educated (Low-EDUC; first quartile), patients with average education (Average-EDUC; second quartile), highly educated (High-EDUC; third quartile), and exceptionally educated (Except-EDUC; fourth quartile). FDG-PET of typical AD, late converters, and aggressive and smoldering AMY+ MCI subgroups, as well as education level-based subgroups, were compared with healthy volunteer control subjects (CTR) and within each group using a two-samples t test design (SPM8; p < 0.05 family-wise error-corrected). Late converters were characterized by relatively preserved metabolism in the right middle temporal gyrus (Brodmann area [BA] 21) and in the left orbitofrontal cortex (BA 47) with respect to typical AD. When compared with CTR, the High-EDUC subgroup demonstrated a more extended bilateral hypometabolism in the posterior parietal cortex, posterior cingulate cortex, and precuneus than the Low- and Average-EDUC subgroups expressing the same level of cognitive impairment. The Except-EDUC subgroup showed a cluster of significant hypometabolism including only the left posterior parietal cortex (larger than the Low- and Average-EDUC subgroups but not further extended with respect to the High-EDUC subgroup). Middle and inferior temporal gyri may represent sites of resilience rather than a hallmark of a more aggressive pattern (when hypometabolic). These findings thus support the existence of a relatively homogeneous AD progression pattern of hypometabolism despite AD heterogeneity and interference of cognitive reserve. In fact, cortical regions whose “metabolic resistance” was associated with slower clinical progression had different localization with respect to the regions affected by education-related reserve.
- 3 April 2018
- Published by Springer Nature in Alzheimer's Research & Therapy
- Vol. 10 (1) , 35
Open AccessMetabolic correlates of reserve and resilience in MCI due to Alzheimer's Disease (AD)
We explored the presence of both reserve and resilience in late-converter mild cognitive impairment due to Alzheimer’s disease (MCI-AD) and in patients with slowly progressing amyloid-positive MCI by assessing the topography and extent of neurodegeneration with respect to both “aggressive” and typically progressing phenotypes and in the whole group of patients with MCI, grounding the stratification on education level. We analyzed 94 patients with MCI-AD followed until conversion to dementia and 39 patients with MCI who had brain amyloidosis (AMY+ MCI), all with available baseline 18F-fluorodeoxyglucose positron emission tomography (FDG-PET) results. Using a data-driven approach based on conversion time, patients with MCI-AD were divided into typical AD and late-converter subgroups. Similarly, on the basis of annual rate of Mini Mental State Examination score reduction, AMY+ MCI group was divided, obtaining smoldering (first tertile) and aggressive (third tertile) subgroups. Finally, we divided the whole group (MCI-AD and AMY+ MCI) according to years of schooling, obtaining four subgroups: poorly educated (Low-EDUC; first quartile), patients with average education (Average-EDUC; second quartile), highly educated (High-EDUC; third quartile), and exceptionally educated (Except-EDUC; fourth quartile). FDG-PET of typical AD, late converters, and aggressive and smoldering AMY+ MCI subgroups, as well as education level-based subgroups, were compared with healthy volunteer control subjects (CTR) and within each group using a two-samples t test design (SPM8; p < 0.05 family-wise error-corrected). Late converters were characterized by relatively preserved metabolism in the right middle temporal gyrus (Brodmann area [BA] 21) and in the left orbitofrontal cortex (BA 47) with respect to typical AD. When compared with CTR, the High-EDUC subgroup demonstrated a more extended bilateral hypometabolism in the posterior parietal cortex, posterior cingulate cortex, and precuneus than the Low- and Average-EDUC subgroups expressing the same level of cognitive impairment. The Except-EDUC subgroup showed a cluster of significant hypometabolism including only the left posterior parietal cortex (larger than the Low- and Average-EDUC subgroups but not further extended with respect to the High-EDUC subgroup). Middle and inferior temporal gyri may represent sites of resilience rather than a hallmark of a more aggressive pattern (when hypometabolic). These findings thus support the existence of a relatively homogeneous AD progression pattern of hypometabolism despite AD heterogeneity and interference of cognitive reserve. In fact, cortical regions whose “metabolic resistance” was associated with slower clinical progression had different localization with respect to the regions affected by education-related reserve.