Lalith kumar Shiyam Sundar

W2 Professor: Digital Transformation in Radiology

Ludwig-Maximilians-Universität München
Country flagGermany

Research Interests

ApplyKite TwinLearns as you apply

Make your next move.

Prepare your application

Review drafts before you send them.

Explore related searches

Articles (17)

Purpose This study aims to investigate whether a diagnostic AI model can effectively support lesion detection and staging in non-small cell lung cancer (NSCLC) [ 1 ⁸F]FDG PET/CT studies, focusing on the distinction between technical segmentation accuracy and clinically meaningful performance. Methods In this retrospective single-centre study, [ 1 ⁸F]FDG PET/CT scans from 306 treatment-naïve NSCLC patients were reviewed with reference to multidisciplinary team decisions. Tumour lesions were manually segmented for reference and compared with predictions from the top-performing algorithm of the autoPET III challenge. Quantitative segmentation metrics were calculated, and lesion-level errors were assessed for impact on patient-level TNM and UICC staging. Results The algorithm achieved a mean Dice Similarity Coefficient (DSC) of 0.64. Lesion-level sensitivity was 95.8% across all patients, with a precision of 87.5%. False positive M-category lesions ( n = 196) occurred as most frequent error. Of all false positives, 35.7% were benign and 34.7% non-oncologic pathologies. UICC staging matched ground truth in 207/306 patients, with most discordances due to upstaging (88/306). Conclusion Clinically driven metrics and cause-based error analysis offer valuable insight into AI segmentation performance. The evaluated model showed excellent lesion sensitivity but a tendency towards systematic overprediction across TNM categories. On a lesion level M-stage false positives and undersegmentation in the hilar region emerged as the main driver of clinically relevant upstaging. Despite promising lesion detection sensitivity, only 67.7% UICC-stagings were accurate using AI masks, indicating that diagnostic AI may support, though not yet replace, manual lesion evaluation in NSCLC [ 1 ⁸F]FDG PET/CT.

Year:

2026

Year:

2026

Year:

2026

Background Manual segmentation of organs from PET/CT images is a time‐consuming and highly operator‐dependent task. Open software solutions are now available to automatically segment all major anatomical structures in CT images. Purpose We compared the volumes and standardized uptake values (SUV) extracted from [18F]FDG‐PET/CT patient scans for 33 anatomical structures segmented using two deep learning (DL) algorithms to determine if they are interchangeable. Methods Baseline [18F]FDG‐PET/CT images were collected retrospectively for 315 women with metastatic breast cancer. A total of 33 anatomical volumes of interest (VOI) were segmented from the whole‐body CT scans using both MOOSE v.3.0.14 and TotalSegmentator v.2.0.5 and copied onto the corresponding PET images. For each VOI, the volume from the CT image and SUVmax, SUVpeak and SUVmean from the PET image were extracted. The resulting values were compared using the relative difference for each feature. Results Following DL segmentation, resulting organ volumes differed by less than 10% for 19/33 organs in more than 80% (252/315) of patients. Four organs were segmented with volume differences greater than 20% in 1/5th of patients: bladder (48%, p < 0.0001), portal and splenic veins (34%, p < 0.0001), thyroid (16%, p < 0.0001), adrenal glands (15%, p < 0.0001). SUVmax and SUVpeak were affected by the choice of DL algorithms, with values differing by less than 10% in more than 80% of patients for only 16 and 19 out of 33 organs, respectively. In contrast, SUVmean was less affected with differences of less than 10% in more than 80% of patients for all anatomical structures, except the bladder, lungs and skull. Conclusions The two software tools produce similar results in volume estimates for most anatomical structures. SUVmean is less dependent on the segmentation algorithm than SUVmax and SUVpeak and shows excellent reproducibility for all anatomical structures studied except for the bladder, the lungs and the skull.

Year:

2025

Background Accurate tumor delineation in somatostatin receptor (SSTR) PET/CT is imperative for quantifying tumor burden, therapy monitoring, and response assessment in gastroenteropancreatic neuroendocrine tumors (GEPNETs). This study aimed to evaluate different SSTR PET-based tumor delineation approaches, based on a lymph node metastasis as reference standard, and determine the most robust method for volume assessment. Singular, non-bulky lymph node metastases with a short axis diameter ≥ 1.0 cm from [¹⁸F]SiTATE PET/CT scans were included. Different tumor delineation methods were applied: fixed SUV thresholds, isocontours relative to SUV max , and individual backwards thresholding. The correlation between PET-derived volumes and CT-derived volumes (reference standard) were investigated. Results An isocontour approach using 45% of SUV max demonstrated the strongest correlation with CT-derived volumes ( r = 0.874; r² = 0.764; p < 0.001), outperforming fixed SUV thresholds, e.g. SUV 4.0 ( r = 0.727; r 2 = 0.529; p < 0.001). The application of a backwards threshold approach resulted in the identification of significant variability (CoV: 59.6% for fixed SUV, 31.3% for isocontour). Conclusion The 45% isocontour-thresholding relative to lesional SUV max approach constitutes a robust and clinically applicable method for SSTR PET-based tumor delineation in GEPNET patients, irrespective of changes of SSTR avidity of reference tissues (e.g. liver). Further validation is required to establish its role in total tumor volume assessment and therapy monitoring.

Year:

2025

Year:

2025

Positron Emission Tomography (PET) with 2-deoxy-2-[ 18 F]fluoro-D-glucose (FDG) and Computed Tomography (CT) are key tools in oncology diagnostics. However, most PET studies focus on tumour detection. We propose that a holistic assessment of the macroenvironment using FDG-PET could enhance our understanding of systemic treatment interactions and improve patient care. To achieve this, we aim to develop a normative database for FDG-PET based on healthy controls, providing a reference for identifying voxel-level metabolic aberrations in cancer patients. This approach may uncover treatment-related changes beyond cancer diagnostics. 48 healthy controls who underwent dynamic test-retest whole-body PET/CT imaging post [ 18 F]FDG injection were included in this cohort. Static PET images (57–62 min post-injection) were reconstructed using CT-based attenuation and scatter correction, with iterative reconstruction incorporating resolution recovery and time-of-flight data. Standardized uptake values (SUVs), tissue densities (HU), and volumes for 135 organs were calculated using in-house segmentation software. The dataset, including anonymized PET/CT images and CT-derived segmentations in NIfTI format, supports the creation of a normative FDG-PET database and enables multi-organ analyses using PET/CT imaging.

Year:

2025

Background [18F]FDG-PET/CT images play a key role in the management of patients with Non-Small Cell Lung Cancer (NSCLC). However, in these scans, the focus is always on the detected tumors and on their characteristics, neglecting information from other organs or tissues. Purpose To investigate whether the mean brain FDG uptake (SUVmean brain ) is associated with overall survival (OS) in metastatic NSCLC patients. Materials and Methods This retrospective study included metastatic NSCLC patients who underwent pre-treatment [18F]FDG-PET/CT scans between 2010 and 2023. Clinical and biological data, tumor radiomic features, and SUVmean brain were collected. The ability of these features to predict OS was evaluated using univariable and multivariable Cox regression models. The correlation between SUVmean brain and clinical/imaging/blood biomarkers was investigated using Spearman correlation coefficients (rS). Results Patients were chronologically divided into a discovery (n = 234, mean age 64 years ± 11 [SD], 135 male) and test set (n = 146, mean age 66 years ± 11, 83 male). In the discovery set, univariable analysis showed that high SUVmean brain (≥ median) was associated with longer OS (HR = 0.83, 95% CI 0.76-0.92, P < .001). SUVmean brain was significantly lower in patients who died within one year compared to those who were still alive ( P < .001). In multivariable analysis, SUVmean brain remained an independent prognostic factor for OS (HR, 0.88; 95% CI: 0.80, 0.98; P = .02), which was confirmed in the test set ( P < .001). SUVmean brain was independent of the radiomic features quantifying tumor involvement (|rS| < 0.24, n = 380 patients) and significantly correlated but complementary to several blood biomarkers including CRP (rS = -0.37, n = 110 patients). Conclusion Low brain metabolic activity was associated with increased mortality in metastatic NSCLC patients. SUVmean brain was an independent prognostic factor that could contribute to patient stratification, although its interpretation requires further investigation.

Year:

2025

Aim: High-sensitivity hybrid positron emission tomography (PET) imaging using advanced whole-body (WB) or total-body PET/computed tomography (CT) systems permits reducing injected tracer activity while preserving diagnostic quality. Such approaches are promising for healthy control studies or exploring inter-organ communication in systemic diseases. This study assessed test/retest variations in the fluoro-2-deoxy-D-glucose (FDG) uptake in key organs from low-dose (LD) and standard-dose (STD) [18F]FDG-PET/CT imaging protocols in healthy controls and lung cancer patients. Methods: A total of 19 healthy controls (19–62 years, 46–104 kg, 10 M/9 F) and 7 lung cancer patients (47–77 years, 50–88 kg, 4 M/3 F) underwent [18F]FDG-PET/CT imaging. All subjects were first injected (“test,” LD) with 28 ± 2 MBq FDG and underwent a dynamic (0–67 min post-injection) WB imaging protocol with LD-CT. Then, 90 min post-LD injection, the subjects were repositioned and injected with 275 ± 16 MBq FDG (“retest,” STD). Second LD-CT and STD-CT scans were acquired for healthy controls and patients, respectively. Static images (55–67 min post-injection) were considered for subsequent analysis. The CT images were used to automatically segment the target volumes of interest. Standardized uptake values normalized to the body weight (SUV BW ) were extracted for each volume of interest. The mean SUV BW were compared for both LD/STD conditions with paired t-tests. In patients, FDG-avid lesions were manually delineated on LD and STD static images. Effective dose levels were estimated from both the CT and PET acquisitions. Results: Organ-based mean SUV BW were similar between the LD and STD (mean %difference ≤5%) in both healthy controls and cancer patients, except in the heart. Intra-control test/retest variability was significant in the brain, heart, and skeletal muscle ( p < 0.05). While 17 lesions were delineated on the STD images of the patients, only 10/17 lesions were identified on the LD images due to increased image noise. Lesion-based mean SUV BW were similar between LD and STD acquisitions ( p = 0.49, %difference = 10%). In patients, the effective doses were (1.9 ± 0.2) mSv (LD-CT), (16.6 ± 5.4) mSv (STD-CT), (0.5 ± 0.1) mSv (LD-PET), and (4.6 ± 0.3) mSv (STD-PET). Conclusion: LD and STD [18F]FDG injections in healthy controls and lung cancer patients yielded comparable mean SUV BW , except in the heart. Dose levels may be reduced for [18F]FDG-PET imaging without a loss in mean SUV BW accuracy, promoting LD-PET/CT protocols for studying multi-organ metabolic patterns. In oncology patients, this approach may be hindered by a lower diagnostic quality in the presence of significant noise.

Year:

2024

Year:

2024

Year:

2024

Year:

2024

Year:

2024

Year:

2023

Year:

2023

Year:

2023

Year:

2023

Collaborators (13)

Thomas Lund

Københavns Universitet

DENMARK
View details

Valentin Riedl

Full Professor "Brain Imaging of Metabolic and Immunological Processes""

Friedrich-Alexander Universität Erlangen-Nürnberg

GERMANY
View details

Sune Hannibal Holm

Københavns Universitet

DENMARK
View details

Yasser Abdelhafez

Specialist

University of California, Davis

UNITED STATES
View details

Ramsey Badawi

Professor

university of california davis medical center

UNITED STATES
View details

Armin Frille

-

GERMANY
View details

Lisa C Adams

Stanford University

UNITED STATES
View details

Gabriel Castrillon

Senior Researcher

Friedrich-Alexander Universität Erlangen-Nürnberg

GERMANY
View details

Peter Sandøe

Professor

University of Copenhagen

DENMARK
View details

Irene Buvat

CNRS Délégation Ile-de-France Sud

FRANCE
View details

Simon R. Cherry

Prof. of Total-Body Imaging

University of California, Davis

UNITED STATES
View details

Nina JEHANNO

Head of Nuclear Medicine

Institut Curie

FRANCE
View details

Ivo Rausch

Medical University of Vienna

AUSTRIA
View details