Background Accurate interpretation of dopamine transporter (DAT)-SPECT depends on clinical context, including symptoms, medication, comorbidities, and structural imaging findings. In routine practice, however, this information is often documented as heterogeneous free text, which may hinder consistent consideration of clinically relevant parameters and limit guideline-compliant reporting. We therefore developed and evaluated a locally deployable small language model (SLM)-based pipeline to structure and standardize the clinical information section for DAT-SPECT reporting. Two retrospective datasets were used. The in-house dataset included clinical information sections from 2,990 DAT-SPECT examinations and was split into training (n = 2,700) and internal test data (n = 290). The external validation dataset comprised 595 clinical information sections from a second institution. A structured dictionary (466 entries across 12 categories) was created by two experienced DAT-SPECT readers and used to manually annotate the training and internal test samples. A Qwen3 1.7B model was fine-tuned by supervised learning to reproduce expert annotations. A rule-based postprocessing pipeline converted model outputs into report-ready structured text, including removal of redundant entries, automated suggestion of scan indication, consolidation of symptom lateralization and time course, reminders for missing guideline-relevant elements, and medication interaction checks. Internal test performance was assessed by expert comparison of SLM output with manual annotation. External performance was assessed by counting the manual corrections required in the postprocessed output under strict criteria. Results In the internal test dataset, the postprocessed SLM output provided the same information as the expert annotation in 196/290 cases (67.6%), including exact matches in 147/290 (50.7%). Interpretation differences due to free-text ambiguity occurred in 13/290 cases (4.5%), and SLM errors in 41/290 cases (14.1%). In 40/290 cases (13.8%), the SLM output was judged superior to the manual annotation. Among the error cases, 36/41 (87.8%) involved only a single erroneous or missing item. In the external dataset, 446/595 postprocessed outputs (75.0%) required no manual correction before inclusion in the report. Among the remaining 149/595 cases (25.0%), correction of only a single item was required in 111 cases (18.7% of all cases). The mean number of corrections was 0.34 items per case, corresponding to 6.0% of all structured items. Conclusions A locally deployable SLM combined with rule-based postprocessing can transform free-text clinical information for DAT-SPECT into standardized, report-ready documentation with high clinical usability and robust external performance. By improving consistency, highlighting missing guideline-relevant information, and enabling real-time support during clinical assessment, the pipeline may facilitate structured reporting without reliance on cloud-based infrastructure.

Aim:This study evaluated ChatGPT (GPT-5.2) for drafting a review paper on deep learning in dopamine transporter (DAT)-SPECT with [¹²³I]ioflupane. Methods:The review workflow consisted of 3 steps: (i) literature search, (ii) generation of structured summaries with 24 predefined fields for each publication, and (iii) drafting a review paper based on the structured summaries across all publications. A detailed prompt for ChatGPT was iteratively designed for each step with ChatGPT support. A manual literature search was independently performed by an expert in DAT-SPECT and deep learning. ChatGPT-generated structured summaries were manually fact-checked against the full publications and corrected where necessary. The review draft generated by ChatGPT was checked against the corrected summaries. Results:When prompted to compile an exhaustive list of publications, ChatGPT cited 13 papers, whereas the manual search identified 70 relevant publications, 67 of which were included. Corrections to ChatGPT-generated structured summaries were required in 27 cases (40.3%), affecting one or two of the 24 predefined fields, while no changes were necessary in 40 publications (59.7%). Most corrections could likely have been avoided by more precise prompting. All numerical information (dataset sizes, train-test splits, performance metrics) was correct. The review draft (~950 words) generated by ChatGPT was content-wise meaningful and accurate, but contained referencing errors, including incorrect citations, missing references, and citations of non-existent publications. Conclusions:ChatGPT is a highly effective tool for drafting review manuscripts in nuclear medicine imaging, but its limitations in literature retrieval and referencing require careful expert supervision.

IMPORTANCE:In critically ill patients with sepsis, the infectious focus may remain unclear despite microbiology, ultrasound, and conventional cross-sectional imaging. In this setting, 18F-Fluorodeoxyglucose Positron Emission Tomography Combined With CT (FDG PET/CT) is used selectively, but its diagnostic value and downstream clinical consequences are not well defined. OBJECTIVES:To assess the diagnostic yield of FDG PET/CT in selected ICU patients with unresolved infection, describe the infectious foci detected, and document subsequent diagnostic and therapeutic actions. DESIGN, SETTING, AND PARTICIPANTS:Retrospective single-center cohort study of 108 FDG PET/CT scans performed in 103 adult patients with sepsis treated in the ICUs of a tertiary university hospital. PET/CT was requested either to search for an unknown infectious focus or to confirm or exclude a suspected infectious focus. MAIN OUTCOMES AND MEASURES:PET/CT findings suspicious for infection, confirmation according to a composite clinical reference standard, classification as clinically informative, diagnostic performance, and diagnostic or therapeutic actions after imaging. RESULTS:FDG PET/CT detected a subsequently confirmed infectious focus in 72 of 108 scans (66.7%). Eight additional findings suspicious for infection were not confirmed, and one scan was classified as false negative. Apparent sensitivity was 98.6% (95% CI, 96.0-100.0%), specificity 77.1% (95% CI, 63.2-91.1%), positive predictive value 90.0% (95% CI, 83.4-96.6%), and negative predictive value 96.4% (95% CI, 89.6-100.0%). When PET/CT was used to search for an unknown focus, at least one infectious site was subsequently confirmed in 53 of 81 scans (65.4%). When it was used to confirm or exclude a suspected focus, results were diagnostically conclusive in 24 of 27 scans (88.9%). Overall, 77 scans (71.3%) were considered clinically informative. Imaging findings were followed by additional noninvasive diagnostics in 50 scans, invasive diagnostic or therapeutic procedures in 52 scans, and antimicrobial modification in 44 scans. CONCLUSIONS AND RELEVANCE:In selected critically ill patients with sepsis and unresolved infection, FDG PET/CT often identified a confirmed infectious focus or helped reduce diagnostic uncertainty. These findings support its use as a complementary, problem-oriented diagnostic tool rather than as routine imaging in ICU sepsis. Because referral was selective, the study was retrospective, and the composite reference standard was imperfect, diagnostic performance estimates should be interpreted cautiously. Prospective studies are needed to define patient selection and determine whether PET/CT improves patient-centered outcomes.

To estimate (i) the rate of difficult-to-interpret cases in dopamine transporter (DAT)-SPECT with [123I]ioflupane and (ii) the diagnostic accuracy of binary visual categorization in difficult-to-interpret DAT-SPECT.The study included 178 control subjects (49% females, 63.7±11.7y) and 178 sex- and age-matched patients with pre-screening clinical diagnosis of Parkinson's disease (PD) from the Parkinson's Progression Markers Initiative (PPMI). SPECT images reconstructed by the PPMI and a local method were visually interpreted twice by 2 independent readers with respect to Parkinson-like reduction of the striatal signal using the following 6-score: -3 = clearly reduced, -2 = probably reduced, -1 = more likely reduced than normal, 1 = more likely normal than reduced, 2 = probably normal, 3 = clearly normal. Cases with 6-score of -1 or 1 were considered "difficult-to-interpret", all other cases were considered "conclusive". To assess diagnostic accuracy relative to the clinical group label (control, PD), the 6-score was binarized using 0 as cutoff.The proportion of difficult-to-interpret cases ranged between 3.4% (95%-CI 1.5-5.2%) and 7.6% (4.8-10.3%) across readers and reconstruction methods. The proportion of cases misclassified by the binarized score ranged between 17.4% (1.9-32.9%) and 50.0% (25.5-74.5%) among the difficult-to-interpret cases, and between 5.1% (2.7-7.4%) and 6.4% (3.8-9.0%) among the conclusive cases.(i) the proportion of difficult-to-interpret cases in DAT-SPECT for the diagnosis of parkinsonism is not larger than 10%, (ii) the diagnostic accuracy of binary decisions is severely reduced in these cases. We therefore recommend reporting difficult-to-interpret cases as "inconclusive", unless in special situations.

BACKGROUND:Multiple-pinhole (MPH) collimators for brain SPECT improve the resolution-sensitivity tradeoff compared with conventional parallel-hole and fan-beam collimators. The system count sensitivity profile with MPH collimators exhibits a peak at the center of the field-of-view and significant decline in count sensitivity toward the periphery. In MPH SPECT with [123I]FP-CIT, this makes the difference in relative statistical noise between the striatum and extrastriatal (reference) regions even bigger. This could slow down the convergence of iterative reconstruction compared with [123I]FP-CIT using conventional collimators. This study evaluated the impact of the effective number of reconstruction iterations in [123I]FP-CIT SPECT with MPH collimators. METHODS:671 patients (43.8% females, 67.3±11.3y) were included retrospectively. Projection data were acquired with a triple-head camera (Mediso AnyScan Trio) equipped with 2nd generation general-purpose brain MPH collimators. MPH projections from 12 min total scan duration were reconstructed with the iterative 3d Tera-Tomo Monte Carlo algorithm of the system software. The effective number of iterations was varied between 12 and 300. To evaluate the impact of the iteration number on diagnostic performance, a data-driven Gaussian mixture model approach was used to assess the power of the putaminal specific binding ratio (SBR) for differentiation between reduced and normal scans. In a subset of 72 patients, reconstructed images were visually evaluated twice by 7 independent readers with respect to parkinson-like reduction of striatal [123I]FP-CIT uptake using a Likert 6-score (-3=clearly reduced, …, 3=clearly normal). The impact of the iteration number on the Likert 6-score was tested using repeated measure analysis of variance (ANOVA) with iteration number, reader and session as within-subject factors. The majority vote binary classification (reduced, normal) was taken into account as between-subject factor. Within- and between-readers agreement of visual scoring were characterized by Cohen's and Fleiss' kappa. RESULTS:Iterative reconstruction converged much slower in the extrastriatal reference region than in the striatum and its subregions. Considerable improvement of spatial resolution with increasing iteration number came with an acceptable increase of statistical noise. The discriminative power of the putaminal SBR was highest with 300 effective iterations. There was a small but significant reduction of reader confidence and between-readers agreement with 300 compared to 24 iterations, particularly in normal cases (Likert 6-score 2.325 versus 2.468, p < 0.001). CONCLUSIONS:300 iterations provide a good compromise between spatial resolution, robustness and diagnostic power of semi-quantitative SBR analyses on one side versus statistical noise and between-readers variability on the other side. In order to avoid relevant loss of between-readers stability, readers should be specifically trained for reading [123I]FP-CIT MPH SPECT images reconstructed with a high number of iterations.

To provide fully automatic scanner-independent 5-level categorization of the [123I]FP-CIT uptake in striatal subregions in dopamine transporter SPECT. A total of 3500 [123I]FP-CIT SPECT scans from two in house (n = 1740, n = 640) and two external (n = 645, n = 475) datasets were used for this study. A convolutional neural network (CNN) was trained for the categorization of the [123I]FP-CIT uptake in unilateral caudate and putamen in both hemispheres according to 5 levels: normal, borderline, moderate reduction, strong reduction, almost missing. Reference standard labels for the network training were created automatically by fitting a Gaussian mixture model to histograms of the specific [123I]FP-CIT binding ratio, separately for caudate and putamen and separately for each dataset. The CNN was trained on a mixed-scanner subsample (n = 1957) and tested on one independent identically distributed (IID, n = 1068) and one out-of-distribution (OOD, n = 475) test dataset. The accuracy of the CNN for the 5-level prediction of the [123I]FP-CIT uptake in caudate/putamen was 80.1/78.0

BACKGROUND:Sentinel lymph node biopsy (SLNB) is an effective method for managing clinically node-negative (cN0) early-stage oral squamous cell carcinoma (OSCC). This study examines lymphatic drainage patterns and clinical outcomes in pT1/T2 OSCC patients undergoing SLNB while establishing optimal thresholds for SLN identification and lymph node removal. METHODS:Clinicopathological data were analyzed to assess associations between SLN localization and tumor characteristics. Logistic regression identified risk factors for contralateral SLN involvement. Receiver operating characteristic curves determined optimal thresholds for SLN identification and lymph node removal. Survival analysis was conducted using Kaplan-Meier curves and log-rank tests. RESULTS:Most SLNs were ipsilateral (83.33 %), with fewer contralateral (8.33 %) or bilateral (8.33 %) cases. The most common SLN sites were Levels 1b and 2a (70.83 %), with Level 2a being the most frequent (39.58 %). Metastases were found in 8.33 % of patients, requiring neck dissection. During follow-up, 6.25 % experienced regional recurrence. Identifying ≥2 SLNs and removing >4 lymph nodes significantly improved recurrence-free survival (p < 0.05). CONCLUSION:SLNB effectively maps lymphatic drainage in small OSCC tumors, detecting atypical pathways and metastases. Identifying at least two SLNs and removing more than four lymph nodes improves recurrence-free survival, highlighting the importance of thorough lymph node removal to optimize patient outcomes and reduce recurrence risk.

Theranostics in nuclear oncology combines diagnostic and therapeutic procedures using radiotracers to target tumor cells. Prostate-specific membrane antigen (PSMA) is a key target in metastatic prostate cancer, and the radioligand [177Lu]Lu-PSMA-617, which binds to PSMA, has shown promising results in treating metastatic castration-resistant prostate cancer (mCRPC), leading to its approval by the European Medicines Agency in 2022. In this narrative review, the current evidence of [177Lu]Lu-PSMA-617 in mCRPC was discussed in the context of selected studies and the joint EANM/SNMMI guidelines for Lutetium-177-labeled PSMA-targeted radioligand therapy. The use of [177Lu]Lu-PSMA-617 for post-chemotherapy mCRPC is supported by substantial evidence from the phase II TheraP and the phase III VISION trials, demonstrating its safety and efficacy. The theranostic approach identifies patients likely to benefit from [177Lu]Lu-PSMA-617, which is effective only in tumors with sufficient PSMA expression, as detected by PSMA-ligand PET/CT, which is also used for response assessment. The success of [177Lu]Lu-PSMA-617 in post-chemotherapy mCRPC patients has led to further ongoing studies evaluating its use earlier in the treatment sequence, prior to chemotherapy. To ensure beneficial treatment outcome, adequate patient selection and evaluation of imaging-based response through PSMA-ligand PET/CT is necessary.

Sentinel lymph node biopsy (SLNB) is a well-established method for managing primary oral squamous cell carcinoma (OSCC), but its role in recurrent or secondary carcinomas, particularly following previous neck dissection, remains insufficiently studied. This study explores the utility of SLNB in these challenging cases, focusing on reliability, sentinel lymph node (SLN) localisation trends, and clinicopathological factors, comparing results between primary and secondary/recurrent carcinomas. A retrospective analysis was conducted on patients with primary and recurrent or secondary OSCCs treated at a German tertiary medical centre. Chi squared tests were used to analyse correlations between clinicopathological characteristics and SLN localisation. The negative predictive value (NPV) was calculated, and the timing of recurrence was also evaluated. We found an NPV of 93.75% for SLNB in primary carcinomas, whereas it was 88.89% for secondary and recurrent carcinomas. While most SLNs were localised ipsilaterally across both groups, recurrent/secondary cancers showed a higher prevalence of contralateral SLNs (22.2%) and bilateral localisation patterns (11.1%) compared to primary carcinomas (8.3% contralateral; 8.3% bilateral). Despite observed trends, no statistically significant associations were found between SLN localisation and clinicopathological factors. The mean (SD) time to recurrence was significantly shorter for primary carcinomas with 9.20 (3.49) months compared to recurrent or secondary cases 32.00 (8.54) months). SLNB in recurrent or secondary OSCC shows distinct SLN localisation patterns, including more contralateral and mixed foci. In terms of NPV, SLNB is reliable in both primary and secondary/recurrent OSCC. Furthermore, a longer recurrence time reinforces the potential of SLNB as a viable alternative to neck dissection in managing these complex cases.

Background/Objectives: To investigate the efficacy and safety of treatment with [177Lu]Lu-PSMA-I&T Radioligand Therapy (PSMA-RLT) in older patients (≥80 years) vs. younger ones with metastatic castration-resistant prostate cancer (mCRPC). Methods: In this retrospective single-center analysis, 103 patients treated with PSMA-RLT between 2019 and 2024 were included. Overall survival (OS) and therapeutic response were assessed by PSA serum and based on PET/CT Imaging according to the RECIP 1.0 criteria, respectively. Toxicity was additionally assessed via laboratory (hemoglobin, cell counts, and serum creatinine). Adverse events (AEs) were detected according to CTCAE V.5. Results: Median OS did not differ significantly in patients ≥ 80 years vs. <80 years (13.7 vs. 16.1 months, respectively). PSA decline of ≥50% was achieved in 32% patients in total, comparably in both groups (29.4% vs. 34.8%). According to RECIP 1.0, the majority of patients with both ≥80 and <80 years demonstrated stable disease or partial responses in imaging (64% and 71%, post two cycles). Concerning toxicity, the most frequently observed AE was anemia, which occurred in both <80 and ≥80 subgroups (grade 3: 2.8% vs. 5.9%); however, no grade 4 anemia was recorded. Renal function remained stable throughout treatment, and no AE grade 3 or higher was observed. Overall, the safety profile was comparable between age groups. Conclusions: Treatment with PSMA-RLT can be both effective and well tolerated in patients with mCRPC aged 80 years and older.

Multiple-pinhole collimators provide considerable improvement of SPECT system count sensitivity. This case report suggests that SPECT with brain-specific multiple-pinhole collimators enables cerebral perfusion imaging with diagnostic quality by an early 12 minutes scan immediately after injection of a standard dose of the dopamine transporter ligand 123 I-FP-CIT. Thus, 123 I-FP-CIT SPECT with multiple-pinhole collimators could assess nigrostriatal degeneration (late image) and extrastriatal involvement (early perfusion image) during the same imaging session. The early image may serve as an alternative to FDG PET in patients with suspicion of an atypical neurodegenerative parkinsonian syndrome. This could streamline diagnostic workflows by reducing the need for additional imaging modalities.

Striatal specific binding ratios (SBR) are widely used to support the interpretation of dopamine transporter SPECT scans. Automatic SBR computation often involves using affine transformations to map the individual SPECT images to an anatomical reference space for ROI analysis using predefined standard masks. This does not account for differences in volumetric scaling between brain structures since, by definition, affine transformations preserve volume ratios. However, striatal volume has been reported to scale proportional to (intracranial volume)0.4, indicating particularly pronounced “negative” allometric scaling. This study aimed to investigate the impact of disregarding allometric scaling on putamen SBR, and to propose an easy-to-implement method to avoid this issue. 656 [123I]FP-CIT SPECT (67.2 ± 11.4y, 44.2

BACKGROUND:Posthepatectomy liver failure (PHLF) remains a life-threatening complication after hepatectomy. To reduce PHLF, a preoperative assessment of liver function is indispensable. For this purpose, 99mTc-mebrofenin hepatobiliary scintigraphy with SPECT (MSPECT) can be used. The aim of the current study was to evaluate the predictive value of MSPECT for PHLF in patients with non-colorectal liver tumors (NCRLT) compared to patients with colorectal liver metastasis (CRLM) undergoing extended liver resection.METHODS:We included all patients undergoing extended liver resections via two-stage procedures between January 2019 and December 2021 at the University Medical Center Hamburg-Eppendorf, Germany. All patients received a preoperative MSPECT.RESULTS:Twenty patients were included. In every fourth patient, PHLF was observed. Four patients had PHLF grade C. There were no differences between patients with CRLM and NCRLT regarding PHLF rate and future liver remnant (FLR) volume. Patients with CRLM had higher mebrofenin uptake in the FLR compared to those with NCRLT (2.49%/min/m2 vs. 1.51%/min/m2; p = 0.004).CONCLUSION:Mebrofenin uptake in patients with NCRLT was lower compared to those patients with CRLM. However, there was no difference in the PHLF rate and FLR volume. Cut-off values for the mebrofenin uptake might need adjustments for different surgical indications, surgical procedures, and underlying diseases.

Aim To investigate the relationship between off-target binding of the amyloid tracer [18F]florbetaben (FBB) in the skull and skull density. Methods Forty-three consecutive patients were included retrospectively (age 70.2±7.5y, 42% females, 65% amyloid-positive). For each patient, CT skull density (in Hounsfield units) and (late) FBB uptake in the skull were obtained using an individual skull mask generated by warping the skull tissue probability map provided by the statistical parametric mapping software package (version SPM12) to the native patient space. Skull FBB uptake (mean of the 10% hottest voxels) was scaled to the individual median FBB uptake in the pons. The association between skull FBB uptake and skull density was tested by correlation analyses. Univariate analysis of variance (ANOVA) of skull FBB uptake with dichotomized skull density (low: ≤ median, high), sex (female, male) and amyloid-status (positive, negative) as between-subjects factors was used to assess the impact of sex and amyloid status. Results There was a significant inverse correlation between skull FBB uptake and skull density (Pearson correlation coefficient -0.518, p < 0.001; Spearman rho -0.321, p = 0.036). The ANOVA confirmed the bone density effect on the FBB uptake in the skull (p = 0.019). In addition, sex (p = 0.012) and density*sex interaction (p = 0.016) had a significant impact. Skull FBB uptake was significantly higher in females with low skull density than for all other combinations of sex and skull density. Amyloid status did not reach statistical significance (p = 0.092). Conclusion Off-target binding of FBB in the skull is inversely associated with skull density. The relationship is mainly driven by females. Amyloid status does not have a major impact on skull FBB binding.

Background and Purpose: Thalamic hypometabolism is a consistent finding in brain PET with F-18 fluorodeoxyglucose (FDG) in patients with neurofibromatosis type 1 (NF1). However, the pathophysiology of this metabolic alteration is unknown. We hypothesized that it might be secondary to disturbance of peripheral input to the thalamus by NF1-characteristic peripheral nerve sheath tumors (PNSTs). To test this hypothesis, we investigated the relationship between thalamic FDG uptake and the number, volume, and localization of PNSTs.Methods: This retrospective study included 22 adult NF1 patients (41% women, 36.2 +/- 13.0 years) referred to whole-body FDG-PET/contrast-enhanced CT for suspected malignant transformation of PNSTs and 22 sex-and age-matched controls. Brain FDG uptake was scaled voxelwise to the individual median uptake in cerebellar gray matter. Bilateral mean and left-right asymmetry of thalamic FDG uptake were determined using a left-right symmetric anatomical thalamus mask. PNSTs were manually segmented in contrast-enhanced CT.Results: Thalamic FDG uptake was reduced in NF1 patients by 2.0 standard deviations (p < .0005) compared to controls. Left-right asymmetry was increased by 1.3 standard deviations (p = .013). Thalamic hypometabolism was higher in NF1 patients with >= 3 PNSTs than in patients with <= 2 PNSTs (2.6 vs. 1.6 standard deviations, p = .032). The impact of the occurrence of paraspinal/paravertebral PNSTs and of the mean PNST volume on thalamic FDG uptake did not reach statistical significance (p = .098 and p = .189). Left-right asymmetry of thalamic FDG uptake was not associated with left-right asymmetry of PNST burden (p = .658).Conclusions: This study provides first evidence of left-right asymmetry of thalamic hypometabolism in NF1 and that it might be mediated by NF1-associated peripheral tumors.

This study examined the impact of venlafaxine and bupropion on the detection of nigrostriatal degeneration by dopamine transporter (DAT)-SPECT. 43 patients (70.7 ± 8.6y, 44

Abstract Background The aim of this study was to assess the impact of the post-injection electrical seizure duration on the identification of the seizure onset zone (SOZ) in ictal brain perfusion SPECT in presurgical evaluation of drug-resistant epilepsy. Methods 176 ictal SPECT performed with 99mTc-HMPAO (n = 140) or -ECD (n = 36) were included retrospectively. Visual interpretation of the SPECT images (together with individual MRI and statistical hyperperfusion maps) with respect to lateralization (right, left, none) and localization (temporal, frontal, parietal, occipital) of the SOZ was performed by 3 independent readers. Between-readers agreement was characterized by Fleiss’ κ. An ictal SPECT was considered "lateralizing" if all readers agreed on right or left hemisphere. It was considered "localizing" if it was lateralizing and all readers agreed on the same lobe within the same hemisphere. The impact of injection latency and post-injection seizure duration on the proportion of lateralizing/localizing SPECT was tested by ANOVA with dichotomized (by the median) injection latency and post-injection seizure duration as between-subjects factors. Results Median [interquartile range] (full range) of injection latency and post-injection seizure duration were 30 [24, 40] (3–120) s and 50 [27, 70] (-20–660) s, respectively. Fleiss’ κ for lateralization of the SOZ was largest for the combination of early (< 30 s) injection and long (> 50 s) post-injection seizure duration (κ = 0.894, all other combinations κ = 0.659–0.734). Regarding Fleiss’ κ for localization of the SOZ in the 141 (80.1%) lateralizing SPECT, it was largest for early injection and short post-injection seizure duration (κ = 0.575, all other combinations κ = 0.329–0.368). The proportion of lateralizing SPECT was lower with short compared to long post-injection seizure duration (estimated marginal means 74.3% versus 86.3%, p = 0.047). The effect was mainly driven by cases with very short post-injection seizure duration ≤ 10 s (53.8% lateralizing). Injection latency in the considered range had no significant impact on the proportion of lateralizing SPECT (p = 0.390). The proportion of localizing SPECT among the lateralizing cases did not depend on injection latency or post-injection seizure duration (p ≥ 0.603). Conclusions Short post-injection seizure duration is associated with a lower proportion of lateralizing cases in ictal brain perfusion SPECT.

This study evaluated the potential to reduce the scan duration in dopamine transporter (DAT) SPECT when using a second-generation multiple-pinhole (MPH) collimator designed for brain SPECT with improved count sensitivity and improved spatial resolution compared with parallel-hole and fanbeam collimators. Methods: The retrospective study included 640 consecutive clinical DAT SPECT studies that had been acquired in list mode with a triple-head SPECT system with MPH collimators and a 30-min net scan duration after injection of 181 ± 10 MBq of [123I]FP-CIT. Raw data corresponding to scan durations of 20, 15, 12, 8, 6, and 4 min were obtained by restricting the events to a proportionally reduced time interval of the list-mode data for each projection angle. SPECT images were reconstructed iteratively with the same parameter settings irrespective of scan duration. The resulting 5,120 SPECT images were assessed for a neurodegeneration-typical reduction in striatal signal by visual assessment, conventional specific binding ratio analysis, and a deep convolutional neural network trained on 30-min scans. Results: Regarding visual interpretation, image quality was considered diagnostic for all 640 patients down to a 12-min scan duration. The proportion of discrepant visual interpretations between 30 and 12 min (1.2%) was not larger than the proportion of discrepant visual interpretations between 2 reading sessions of the same reader at a 30-min scan duration (1.5%). Agreement with the putamen specific binding ratio from the 30-min images was better than expected for 5% test-retest variability down to a 10-min scan duration. A relevant change in convolutional neural network-based automatic classification was observed at a 6-min scan duration or less. Conclusion: The triple-head SPECT system with MPH collimators allows reliable DAT SPECT after administration of about 180 MBq of [123I]FP-CIT with a 12-min scan duration.

Abstract Background Ictal brain perfusion SPECT provides higher sensitivity for the identification of the epileptic seizure onset zone (SOZ) than interictal SPECT. However, ictal SPECT is demanding due to the unpredictable waiting period for the next seizure to allow for ictal tracer injection. Thus, starting with an interictal scan and skipping the ictal scan if the interictal scan provides a SOZ candidate with high confidence could be an efficient approach. The current study estimated the rate of high-confidence SOZ candidates and the false lateralization rate among them for interictal and ictal SPECT. Methods 177 patients (48% females, median age 38y, interquartile range 27–48y) with ictal and interictal SPECT acquired with 99mTc-HMPAO (n = 141) or -ECD (n = 36) were included retrospectively. The vast majority of the patients was suspected to have temporal lobe epilepsy. Visual interpretation of the SPECT data was performed independently by 3 readers in 3 settings: “interictal only” (interictal SPECT and statistical hypoperfusion map), “ictal only” (ictal SPECT and hyperperfusion map), and “full” setting (side-by-side interpretation of ictal and interictal SPECT including statistical maps and SISCOM analysis). The readers lateralized the SOZ (right, left, none) and characterized their confidence using a 5-score. A case was considered "lateralizing with high confidence” if all readers lateralized to the same hemisphere with at least 4 of 5 confidence points. Lateralization of the SOZ in the “full” setting was used as reference standard. Results The proportion of “lateralizing with high confidence” cases was 4.5/31.6/38.4% in the “interictal only”/“ictal only”/“full” setting. One (12.5%) of the 8 cases that were “lateralizing with high confidence” in the “interictal only” setting lateralized to the wrong hemisphere. Among the 56 cases that were “lateralizing with high confidence” in the “ictal only” setting, 54 (96.4%) were also lateralizing in the “full” setting, all to the same hemisphere. Conclusions Starting brain perfusion SPECT in the presurgical evaluation of epilepsy with an interictal scan to skip the ictal scan in case of a high-confidence interictal SOZ candidate is not a useful approach. In contrast, starting with an ictal scan to skip the interictal scan in case of a high-confidence ictal SOZ candidate can be recommended.

Deep convolutional neural networks (CNN) hold promise for assisting the interpretation of dopamine transporter (DAT)-SPECT. For improved communication of uncertainty to the user it is crucial to reliably discriminate certain from inconclusive cases that might be misclassified by strict application of a predefined decision threshold on the CNN output. This study tested two methods to incorporate existing label uncertainty during the training to improve the utility of the CNN sigmoid output for this task. Three datasets were used retrospectively: a “development” dataset (n = 1740) for CNN training, validation and testing, two independent out-of-distribution datasets (n = 640, 645) for testing only. In the development dataset, binary classification based on visual inspection was performed carefully by three well-trained readers. A ResNet-18 architecture was trained for binary classification of DAT-SPECT using either a randomly selected vote (“random vote training”, RVT), the proportion of “reduced” votes ( “average vote training”, AVT) or the majority vote (MVT) across the three readers as reference standard. Balanced accuracy was computed separately for “inconclusive” sigmoid outputs (within a predefined interval around the 0.5 decision threshold) and for “certain” (non-inconclusive) sigmoid outputs. The proportion of “inconclusive” test cases that had to be accepted to achieve a given balanced accuracy in the “certain” test case was lower with RVT and AVT than with MVT in all datasets (e.g., 1.9