
BACKGROUND: Deep-learning applications in nuclear medicine are expanding, but many departments lack dedicated AI engineering support, creating a gap between interest and practical implementation. The aim of this study was to evaluate the feasibility of a physician-led, incremental workflow for developing reproducible deep-learning prototypes for nuclear medicine imaging across increasing data complexity (2D planar, sequential planar “3D-like”, and true 3D DICOM) and to quantify prototype discrimination where reference standards were available. MATERIAL AND METHODS: Models were developed in Google Colab Pro using TensorFlow/Keras in a three-step framework. Step 1 used exported planar whole-body scintigraphy images (287 bone scintigraphy; 256 besilesomab scintigraphy) for radiopharmaceutical discrimination and an exploratory positive/negative subclassification without an external reference standard. Step 2 used ordered planar V/Q image stacks from 287 patients; reports provided reference-standard labels for acute pulmonary embolism (perfusion) and ventilation defects. Step 3 used true 3D PET DICOM volumes (46 [¹⁸F]FDG, 42 [¹⁸F]FDOPA, 42 [¹⁸F]FCH) with report-derived radiopharmaceutical labels. Datasets were split into 80% training and 20% validation subsets; performance was reported on the held-out validation subset unless otherwise stated. Primary outcome was feasibility (end-to-end pipeline completion and reproducible inference from an exported model); secondary outcomes included accuracy and AUC (plus sensitivity/specificity for Step 2). RESULTS: Feasibility criteria were met across all three steps, with trained models exported and reloaded for inference in a separate application context. Step 1 radiopharmaceutical discrimination was successful in a small illustrative evaluation set (n = 20 images), while positive/negative subclassification demonstrated reduced reliability consistent with label ambiguity (exploratory). In Step 2, the perfusion model achieved an accuracy of 0.85 and an AUC of 0.96, with a sensitivity of 96.6% and a specificity of 86.6%; the ventilation model achieved an accuracy of 0.82 and an AUC of 0.93, with a sensitivity of 88.4% and a specificity of 76.9%. In Step 3, the 3D ResNet34 achieved a validation accuracy of 1.00 and an AUC of 0.95 for radiopharmaceutical discrimination. CONCLUSIONS: This study demonstrates that deep-learning-based AI can be feasibly implemented even in small nuclear medicine departments. Although prototype in nature, it outlines a complete workflow from data extraction to model training and inference. Crucially, it highlights key limitations — such as reference-standard quality, task definition, preprocessing sensitivity, and the absence of an independent test set — that must be addressed to obtain a reliable and clinically meaningful in-house model.
BACKGROUND:The clinical diagnostic accuracy of idiopathic Parkinson's disease (IPD) remains suboptimal, with the most common neuropathological findings in misdiagnosed patients being atypical neurodegenerative parkinsonism - progressive supranuclear palsy (PSP), multiple system atrophy (MSA), or corticobasal degeneration (CBD). The purpose of this study was to evaluate cardiac mIBG scintigraphy as an imaging biomarker in the differential diagnosis of IPD. MATERIAL AND METHODS:A total of 129 patients with parkinsonism were referred to Center for Nuclear Medicine with PET, University Clinical Center of Serbia, for evaluation of cardiac sympathetic innervation from January 2012 to February 2025. Planar chest scintigrams were obtained 15 min and 4 h after [¹²³I]I-mIBG administration, and early and late heart-to-mediastinum ratio (HMRE and HMRL) and washout rate (WR) were calculated. The final diagnosis was made by a movement disorder specialist. RESULTS:Final diagnosis was available in 103 patients, with IPD in 52 cases, atypical parkinsonism (MSA, PSP or CBD) in 35 patients, and other parkinsonism forms in 16 cases. The areas under the curve (AUCs) for correctly differentiating IPD from other forms of parkinsonism for HMRE, HMRL and WR were 0.769 (95% CI 0.676-0.862), 0.842 (95% CI 0.763-0.921) and 0.748 (95% CI 0.652-0.845), respectively. Sensitivity, specificity, and accuracy of HMRL at a cut-off value of 1.875 were 84.6%, 78.4% and 81.6%, and positive predictive value (PPV) and negative predictive value (NPV) 80% and 83.3%. The AUC for differentiating IPD from MSA using HMRL was 0.829 (95% CI 0.725-0.934), with specificity of 75% for HMRL at cut-off of 1.875. CONCLUSIONS:Cardiac [¹²³I]I-mIBG scintigraphy represents a useful biomarker for differentiating IPD from other forms of parkinsonism, showing high sensitivity and specificity at a cut-off value of 1.875 for late heart-to-mediastinum ratio in our group.
BACKGROUND:Accurate measurement of the glomerular filtration rate (GFR) is essential for the assessment of kidney function and staging of chronic kidney disease (CKD). Renal scintigraphy using technetium-99m diethylenetriamine pentaacetic acid ([99mTc]Tc-DTPA) is widely used for camera-based GFR estimation. However, extravasation of the radiotracer at the injection site may reduce the systemic dose and underestimate GFR. This study evaluated the impact of [99mTc]Tc-DTPA extravasation on GFR estimation and antecubital fossa count-based dose correction on Gates-derived GFR. MATERIAL AND METHODS:An observational cohort study of 50 patients undergoing [99mTc]Tc-DTPA renal scintigraphy assessed injection-site extravasation using antecubital fossa static imaging. GFR was calculated using the Gates method with and without antecubital count-based dose correction. Paired t-tests were used to compare corrected and uncorrected GFR values. Agreement between the two measurements was evaluated using Bland-Altman analysis. Clinical reclassification at the GFR threshold of 60 mL/min was assessed using McNemar's test. RESULTS:The mean corrected GFR was 61.56 ± 19.16 mL/min, compared to 58.98 ± 19.16 mL/min for the uncorrected GFR, demonstrating a statistically significant mean difference of 2.58 mL/min (p < 0.0001; Cohen's dz = 2.07). Bland-Altman analysis showed a mean bias of 2.58 mL/min, and the estimated GFR progressively decreased, with increasing severity of radiotracer extravasation. CONCLUSIONS:Injection-site extravasation results in systematic underestimation of Gates-derived GFR. Antecubital count-based dose correction increases the estimated GFR and may influence clinical interpretation in borderline cases. Routine assessment of injection-site extravasation may improve the accuracy and reliability of camera-based GFR estimation in renal scintigraphy.
BACKGROUND:Accurate staging and metabolic characterization of cervical cancer remains challenging despite advances in imaging. This study investigates the diagnostic utility of dual-phase [18F]FDG PET/CT in cervical carcinoma, with an emphasis on metabolic biomarkers and prognostic indicators. MATERIAL AND METHODS:A prospective study of 50 patients with histopathologically proven cervical cancer was conducted over 24 months. Patients underwent dual-phase [18F]FDG PET/CT imaging with initial whole-body acquisition at 40-60 minutes post-injection and delayed imaging at 120 minutes post-injection following intravenous furosemide administration. Standardized uptake values (SUV), lesion detection efficacy, lymph node characterization, and extra-pelvic metastatic patterns were analyzed using paired t-tests and chi-square analysis. RESULTS:Mean SUV in initial images (9.074 ± 6.113) increased significantly to 12.964 ± 7.494 in delayed images (p < 0.001), representing a 42.8% increase in metabolic activity. Delayed imaging detected primary lesions in 48 cases (96%) versus 44 cases (88%) on initial imaging, demonstrating superior sensitivity (91.7%) and specificity (100%). Critically, seven patients (14%) demonstrated lymph node visualization only on delayed imaging, and 14% exhibited SUV kinetics consistent with high-grade malignancy. Parametrial involvement occurred in 66% of cases, with 81.8% demonstrating bilateral involvement. Extra-pelvic metastases including supraclavicular lymphadenopathy (10%), hepatic involvement (10%), and skeletal lesions (8%) were detected with superior specificity compared to conventional modalities. This study identified a novel metabolic phenotype: "delayed-accumulator lesions" (DAL), characterized by significant SUV escalation between initial and delayed imaging, associated with aggressive biological behavior and poor prognostic outcomes. Additionally, a mathematical model incorporating dual-phase SUV kinetics, parametrial invasion status, and bilateral lymph node involvement demonstrated superior prognostic stratification (AUC > 0.92) compared to conventional FIGO staging. CONCLUSIONS:Dual-phase [18F]FDG PET/CT provides significantly enhanced metabolic characterization of cervical cancer, identifies novel biological markers predictive of treatment response, and enables metabolically guided precision treatment planning. The identification of delayed-accumulator lesions and dual-phase kinetic modeling represent paradigm shifts in cervical cancer stratification, with the potential to revolutionize treatment intensity selection and surveillance protocols.
BACKGROUND:Neuroblastoma (NB) is the most common extracranial solid tumor in children. Metaiodobenzylguanidine (MIBG) selectively accumulates in adrenergic tissue, enabling its use for diagnostic imaging of NB. The implementation of new SPECT/CT protocols allows for the application of novel imaging parameters that may improve diagnostic accuracy and disease monitoring. This study aimed to evaluate the usefulness of AC Quant protocols and parameters in the diagnosis of NB. MATERIAL AND METHODS:This retrospective analysis included 16 treatment-naive children with NB who underwent SPECT/CT with [¹²³I]I-MIBG. AC Quant parameters of the primary tumor and selected organs were analyzed statistically. In the study group, the following tumor parameters were assessed: volume (mL), Hounsfield units (HU), standard deviation of HU (SDHU), SUVmax, SUVpeak, and metabolic tumor volume (MTV). Additionally, SUVmax values for the pancreas, liver, aorta, spleen, and skeletal muscles were measured. RESULTS:Eight of the sixteen patients underwent chemotherapy. Tumor SUVmax and SUVpeak did not change significantly after treatment (p = 0.068). Chemotherapy resulted in a significant increase in SDHU (p = 0.027) and a significant decrease in MTV (p = 0.043). SUV values in the pancreas, liver, aorta, and spleen increased after chemotherapy (p = 0.010; p = 0.007; p = 0.029; p = 0.050, respectively). Tumor-to-pancreas, tumor-to-liver, tumor-to-aorta, tumor-to-spleen, and tumor-to-muscle ratios did not change significantly (p = 0.326; p = 0.176; p = 0.944; p = 0.674; p = 0.484, respectively), with the smallest differences observed for the tumor-to-aorta and tumor-to-muscle ratios. CONCLUSIONS:AC Quant parameters demonstrate potential utility in the assessment of patients with NB. Chemotherapy significantly reduces MTV and increases tumor density heterogeneity in NB. The aorta and paraspinal muscles appear to be the most suitable reference organs for quantitative assessment in NB imaging.
Background: The current standard method in hyperparathyroidism imaging is the combination of[99mTc]Tc-methoxyisobutylisonitrile ([99mTc]Tc-MIBI) single photon emission computed tomography/computed tomography (SPECT/CT) and neck ultrasonography (nUS). However, [99mTc]Tc-MIBI scintigraphy has certain limitations; thus, alternative imaging techniques may be beneficial. The following paper discusses the usefulness of performing an additional 99mTc-pertechnetate ([99mTc]NaTcO4) SPECT/CT after a standard combination of [99mTc]Tc-MIBI SPECT/CT and neck ultrasonography to improve the efficiency of detecting parathyroid adenomas and/or hyperplasia in inconclusive cases. Material and methods: The authors retrospectively analyzed archived medical data of 58 patients who underwent parathyroid scintigraphy in 2022-2024. The study consists of patients, in whose cases [99mTc]Tc-MIBI SPECT/CT in combination with nUS did not provide a clear outcome. These patients underwent additional [99mTc]NaTcO4 SPECT/CT imaging. Radiotracer uptake in both studies was then compared visually. Results: Out of 58 patients included in the study, 40 (69%) had positive [99mTc]Tc-MIBI and [99mTc]NaTcO4 SPECT/CT results, whereas 18 patients (31%) had a negative scintigraphy result. In 42 out of 58 cases, clinical verification of SPECT/CT results was obtained, based on which the sensitivity, specificity, and accuracy of the method were 83.3%. Conclusion: This study demonstrates that additional [99mTc]NaTcO4 SPECT/CT can be valuable in patients with inconclusive results in [99mTc]Tc-MIBI SPECT/CT (evaluated in combination with nUS), both to confirm and exclude the presence of hyper-functioning parathyroid glands.
BACKGROUND:The current standard method in hyperparathyroidism imaging is the combination of [99mTc]Tc-methoxyisobutylisonitrile ([99mTc]Tc-MIBI) single photon emission computed tomography/computed tomography (SPECT/CT) and neck ultrasonography (nUS). However, [99mTc]Tc-MIBI scintigraphy has certain limitations; thus, alternative imaging techniques may be beneficial. The following paper discusses the usefulness of performing an additional 99mTc-pertechnetate ([99mTc]NaTcO₄) SPECT/CT after a standard combination of [99mTc]Tc-MIBI SPECT/CT and neck ultrasonography to improve the efficiency of detecting parathyroid adenomas and/or hyperplasia in inconclusive cases. MATERIAL AND METHODS:The authors retrospectively analyzed archived medical data of 58 patients who underwent parathyroid scintigraphy in 2022-2024. The study consists of patients, in whose cases [99mTc]Tc-MIBI SPECT/CT in combination with nUS did not provide a clear outcome. These patients underwent additional [99mTc]NaTcO4 SPECT/CT imaging. Radiotracer uptake in both studies was then compared visually. RESULTS:Out of 58 patients included in the study, 40 (69%) had positive [99mTc]Tc-MIBI and [99mTc]NaTcO₄ SPECT/CT results, whereas 18 patients (31%) had a negative scintigraphy result. In 42 out of 58 cases, clinical verification of SPECT/CT results was obtained, based on which the sensitivity, specificity, and accuracy of the method were 83.3%. CONCLUSIONS:This study demonstrates that additional [99mTc]NaTcO₄ SPECT/CT can be valuable in patients with inconclusive results in [99mTc]Tc-MIBI SPECT/CT (evaluated in combination with nUS), both to confirm and exclude the presence of hyperfunctioning parathyroid glands.
This research reports a case involving a 53-year-old man with a history of two renal transplants, each placed in one of the lower abdominal quadrants nine years apart, who underwent [99mTc]Tc-ethylene dicysteine (EC) dynamic scintigraphy to assess renal function and hydroureteronephrosis. The scintigraphy revealed an unexpected area of increased extra-renal tracer uptake in the lower pelvic region during both blood flow and functional phases of the study. Subsequent single-photon emission computed tomography/computed tomography (SPECT/CT) images localized this uptake to a fracture in the right pubic bone. This case highlights the necessity of recognizing atypical [99mTc]Tc-EC uptake patterns, which can provide critical clinical insights for the interpretation of unusual imaging findings and enhance diagnostic accuracy.
This case describes a 16-year-old female with papillary thyroid carcinoma who demonstrated persistent iodine uptake in the mediastinum and a pulmonary lesion on post-therapeutic scintigraphy despite an excellent biochemical response. Although fluorine-18 fluorodeoxyglucose positron emission tomography/computed tomography ([¹⁸F]FDG PET/CT) suggested a benign process, persistent uptake and rising anti-thyroglobulin antibody levels prompted repeat radioiodine therapy and thoracotomy. Histopathological examination revealed a bronchogenic cyst and a vascular malformation, with no evidence of metastatic disease. This case highlights the diagnostic challenge posed by benign thoracic lesions exhibiting iodine uptake. Recognition of such false-positive findings is essential to avoid overtreatment in patients with thyroid cancer.
INTRODUCTION: Bone scintigraphy (BS) has emerged as a non-invasive technique of increasing importance in the diagnosis of transthyretin cardiac amyloidosis (CA). The most commonly used approaches include visual grading and semi-quantitative analysis. However, these techniques are limited by operator dependence and subjectivity in interpretation. To address these challenges, absolute quantification techniques are being explored to enhance diagnostic accuracy and consistency and to minimize interobserver variability. The objective is to identify the quantification methods currently employed in the assessment of CA. MATERIAL AND METHODS: A systematic review was conducted, including 12 articles retrieved from Scopus, PubMed, and Web of Science databases. Studies published in the last 5 years were selected, focusing on quantification methods applied to planar imaging and single photon emission computed tomography combined with computed tomography (SPECT/CT) in the context of CA. RESULTS: All 12 selected studies (100%) utilized visual grading; semi-quantitative methods were reported in 91.7% of studies, while absolute quantification techniques were applied in 33.3%. Among the semi-quantitative methods, the heart-to-contralateral lung (H/CL) ratio was the most commonly reported, with similar cut-off values (± 1.5) in the different articles. Additional ratios [heart-to-rib (H/R), heart-to-whole-body (H/WB), heart-to-pelvis (H/P), ratio of heart-to-thigh (RHT)] appeared less frequently, though they were investigated as alternative diagnostic markers. Absolute quantification methods reported heterogeneous cut- -off thresholds, ranging from 1.25 to 6.1, based on standardized uptake values (SUVs), to discriminate between individuals with and without CA. CONCLUSIONS: Visual grading remains the clinical standard approach. The H/CL ratio is the most reported semi-quantitative method, although it presents some limitations. Absolute quantification with SPECT/CT (SUVmax, SUVpeak) is promising for diagnosis, prognosis, and monitoring of transthyretin amyloidosis (ATTR) CA, yet consensus reference values are lacking.
BACKGROUND:Renal scintigraphy using the 99m technetium-diethylene triamine pentaacetic acid [99mTc]Tc-DTPA Gates' protocol is widely employed for glomerular filtration rate (GFR) estimation. However, its accuracy is significantly influenced by the method used for renal depth estimation. This study systematically compared GFR estimation by Gates' protocol using three different depth correction techniques with the gold standard plasma sample analysis method. MATERIAL AND METHODS:[99mTc]Tc-DTPA renal scintigraphy was done as per standard protocol in 40 voluntary renal donors. Renal depth obtained by Tonnesen formula, lateral depth and computed tomography (CT) scan methods. Double plasma sample analysis was done simultaneously. GFR was calculated by Gates' protocol using the three methods of kidney depth measurement separately. RESULTS:The mean total GFR, when measured using the gold standard plasma sample analysis, lateral depth, Tonnesen formula and CT scan, were 83.3 ± 22.34 mL/min, 90.7 ± 21.00 mL/min, 72.1 ± 19.85 mL/min and 89.8 ± 22.55 mL/min, respectively. The Pearson correlation coefficient demonstrated a positive correlation between plasma sample analysis and the 3 depth correction methods - lateral depth method (r = 0.524), Tonnesen formula (r = 0.697), and CT scan (r = 0.597). The intraclass correlation coefficient (ICC) - lateral depth method (ICC = 0.667), Tonnesen formula (ICC = 0.759), and CT method (ICC = 0.733) indicated varying levels of reliability. The Bland-Altman plot analysis revealed poor agreement - lateral depth technique (p = 0.033), Tonnesen formula (p < 0.001), and CT method (p = 0.049). Tonnesen formula exhibited a narrower distribution compared to the other two methods, with a spread of agreement of 64.97. CONCLUSIONS:The findings indicate that, while all three methods exhibited statistically significant variations from plasma sample-derived GFR, the Tonnesen formula showed the closest agreement, albeit with a tendency to underestimate GFR. In contrast, the lateral depth method and CT-derived renal depth consistently overestimated GFR, emphasising the variability introduced by different depth estimation techniques.
Castleman's disease is a rare lymphoproliferative disorder that can mimic malignancy in cancer patients. This study presents a rare case of Castleman disease coexisting with prostate cancer, highlighting the potential value of gallium 68-labelled prostate-specific membrane antigen-11 ([⁶⁸Ga]Ga-PSMA-11) positron emission tomography/computed tomography (PET/CT) in distinguishing Castleman disease from prostate cancer metastasis.