Syahril Siregar
Department Of Physics, Faculty Of Mathematics And Natural Sciences, Universitas Indonesia, Depok, 16424

Published : 2 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 2 Documents
Search

Impact of Training Dataset Size on the Accuracy of L-SVR Single-Time-Point Renal Dosimetry for [¹⁷⁷Lu]Lu-PSMA-617 Therapy Abdurrahman Aziz Wicaksono; Jaja Muhammad Jabar; Syahril Siregar; Deni Hardiansyah
BULETIN FISIKA Vol. 27 No. 1 (2026): BULETIN FISIKA
Publisher : Departement of Physics Faculty of Mathematics and Natural Sciences, and Institute of Research and Community Services Udayana University, Kampus Bukit Jimbaran Badung Bali

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.24843/BF.2026.v27.i01.p07

Abstract

Radiopharmaceutical therapy (RPT) using [¹⁷⁷Lu]Lu-PSMA-617 requires accurate dosimetry to evaluate organs-at-risk (OAR), specifically the kidneys. Single-time-point (STP) dosimetry simplifies clinical workflows by reducing SPECT/CT acquisition. Machine learning (ML) offers a potential solution, yet clinical implementation is hindered by the scarcity of sufficient training datasets for ML-based studies. This study investigated the relationship between training dataset size and time-integrated activity (TIA) estimation accuracy. A Linear Support Vector Regression (L-SVR) model was trained on synthetic virtual patients (VPs, 5,000 total) simulated from a published PBMS NLMEM renal biokinetics at five imaging times (t=1.8 h, 18.7 h, 42.6 h, 66.2 h, and 160.3 h). Time-activity-curve (TAC) and reference TIA (rTIA) were calculated for each VP. Random sampling was performed in increasing dataset sizes. Sample sizes were sub-sampled to training (80%) and testing (20%) datasets. L-SVR was trained on STP data at 42.6 h post-injection (best-time-point of PBMS NLMEM study) from the training dataset and tested by generating estimated TIA (eTIA) with input from the testing dataset. Performance was evaluated by calculating root-mean-square-error (RMSE) and mean-absolute-percentage-error (MAPE) of the eTIA to rTIA. Results showed that the accuracy of eTIA from ML STP dosimetry depends on training size: small samples (n=10) yielded poor performance (RMSE>85.98%, MAPE>89.1%). Accuracy improved significantly at n=500 (RMSE=14.07%) and plateaued beyond n=1,000 (peak RMSE=13.07%). Results indicate that the L-SVR model of the study requires sample sizes of n>200, with optimal gains up to n=2,000. This study suggests synthetic data as a methodological bridge between limited clinical datasets and data-intensive ML approaches.
Hierarchical Tissue-Based MRI Features with Explainable Machine Learning for Alzheimer’s Disease Classification Muhammed B Ceesay; Adhi Harmoko Saputro; Syahril Siregar
Jurnal Ilmu Fisika Vol 18 No 1 (2026): March 2026
Publisher : Jurusan Fisika FMIPA Universitas Andalas

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.25077/jif.18.1.93-104.2026

Abstract

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder characterized by multiscale structural brain degeneration. Many MRI-based machine learning approaches rely on coarse volumetric measures or black-box models with limited anatomical interpretability. This study aims to localize anatomically meaningful brain regions that discriminate AD from cognitively normal (CN) subjects using a hierarchical tissue-based (HTB) MRI framework. The method models gray matter (GM), white matter (WM), and cerebrospinal fluid (CSF) volumetric changes at lobar, gyral, and 246 fine-grained subregions defined by the Brainnetome atlas. T1-weighted MRI scans from 454 participants (227 AD, 227 CN) obtained from ADNI and MIRIAD were preprocessed using AC-PC alignment, N4 bias correction, skull stripping, and nonlinear registration to MNI space. A total of 561 HTB features were extracted to train Random Forest and XGBoost classifiers using five-fold stratified cross-validation with Bayesian hyperparameter optimization. The XGBoost model achieved the best performance (Accuracy: 79.74%, ROC-AUC: 85.07%), comparable to recent atlas-based MRI classification studies, while providing improved multiscale anatomical interpretability. SHAP analysis revealed consistent hierarchical atrophy patterns in hippocampal subregions, medial amygdala, and areas 35/36 and 28/34, demonstrating that hierarchical anatomical modeling with explainable machine learning enables transparent localization of clinically meaningful AD biomarkers without reliance on black-box architectures.