ScholarWorks is an archive that shares over 50,000 articles, working papers, chapters, presentations, posters, theses, historical documents and other items submitted by members of the IU Indianapolis campus community.
Get started! Login with your IU credentials to share freely with 2 million readers per year.
Communities in ScholarWorks
Select a community to browse its collections.
Recent Submissions
GlioMODA: Robust glioma segmentation in clinical routine
(Oxford University Press, 2026-02-12) Canisius, Julian; Buchner, Josef; Rosier, Marcel; Griessmair, Michael; Peeken, Jan C.; Kirschke, Jan S.; Piraud, Marie; Bakas, Spyridon; Menze, Bjoern; Wiestler, Benedikt; Kofler, Florian; Pathology and Laboratory Medicine, School of Medicine
Background: Precise glioma segmentation in magnetic resonance imaging (MRI) is essential for accurate diagnosis, optimal treatment planning, and advancing clinical research. However, most deep learning approaches require complete, standardized MRI protocols that are frequently unavailable in routine clinical practice. This study presents and evaluates GlioMODA, a robust deep learning framework designed for automated glioma segmentation that delivers consistent high performance across varied and incomplete MRI protocols.
Methods: GlioMODA was trained and validated on the BraTS 2021 dataset (1251 training, 219 testing cases), systematically assessing performance across 11 clinically relevant MRI protocol combinations. Segmentation accuracy was evaluated using Dice similarity coefficients (DSC) and panoptic quality metrics. Volumetric accuracy was benchmarked against manual ground truth, and statistical significance was established via Wilcoxon signed‑rank tests with Benjamini-Yekutieli correction.
Results: GlioMODA demonstrated state-of-the-art segmentation accuracy across tumor subregions, maintaining robust performance with incomplete or heterogeneous MRI protocols. Protocols including both T1-weighted contrast-enhanced and T2-FLAIR sequences yielded volumetric differences vs manual ground truth that were not statistically significant for enhancing tumor (median difference 55 mm³, P = .157) and whole tumor (median difference -7 mm³, P = 1.0), and exhibited median DSC differences close to zero relative to the 4‑sequence reference protocol. Omitting either sequence led to substantial and significant volumetric errors.
Conclusions: GlioMODA facilitates reliable, automated glioma segmentation using a streamlined 2‑sequence protocol (T1‑contrast + T2‑FLAIR), supporting clinical workflow optimization and broader implementation of quantitative volumetry compatible with RANO 2.0 criteria. GlioMODA is published as an open-source, easy-to-use Python package at https://github.com/BrainLesion/GlioMODA/.
Approaching Wicked Problems with Design Facilitation & Co-Design: A Case Study Envisioning Third-Party Logistics for the Food Pantry Ecosystem
(2025-04) Perry-Goldsmith, Amy; Hong, Youngbok; Ganci, Aaron; Chernicoff, William
Food insecurity has become a serious concern for many communities in the United States since the COVID-19
pandemic. Embedded challenges in the food pantry ecosystem, including inadequate resources and a largely
volunteer workforce, make the logistics of food distribution inefficient.
Multi-faceted societal problems like food insecurity and food pantry logistics that are ultimately too complex to
be truly solved are known in the design world as “wicked problems.” Designers can approach these problems
with tools like design facilitation and co-design to engage stakeholders, gain insights from diverse groups of people,
and help the community generate authentic potential solutions.
Stakeholders in the Indianapolis food pantry community participated in a designer-facilitated co-design workshop
that led to the creation of three prototypes for next generation logistics models for the food pantry system. Now,
these prototypes can potentially be used by the community to test new ways of distributing food and to seek funding
for future support as changes are made.
Understanding the Use of Artificial Intelligence in Design Curricula
(2025-05) Vavhal, Ria; Ganci, Aaron; Maruf Mihçi, Gürkan; Datta, Amrita
Artificial Intelligence (AI) is rapidly transforming creative industries, yet graphic design education remains
underprepared to equip students with the necessary skills to adapt. This thesis explores how AI can be
meaningfully integrated into design curricula through qualitative research involving interviews with educators
and focus groups with students. The findings revealed diverse perspectives on AI and emphasized the need
for both technical literacy and soft skills such as critical thinking and ethical awareness. In response, this
study developed The Beginner’s Guide to AI in Graphic Design Education—a practical, educator-informed
resource designed to support responsible AI use in the classroom. The guide was evaluated by design faculty
and aims to assist educators in confidently adapting their teaching to an evolving technological landscape.
Detecting Hemorrhagic Myocardial Infarction With 3.0-T CMR: Insights Into Spatial Manifestation, Time-Dependence, and Optimal Acquisitions
(Elsevier, 2025) Chen, Yinyin; Jin, Hang; Guan, Xingming; Yang, Hsin-Jung; Zhang, Xinheng; Chen, Zhenhui; Chan, Shing Fai; Singh, Dhirendra; Jambunathan, Nithya; Youssef, Khalid; Vora, Keyur P.; Gruionu, Gabriel; Dharmakumar, Sanjana K.; Schmeisser, Glen; Tang, Richard; Zeng, Mengsu; Dharmakumar, Rohan; Radiology and Imaging Sciences, School of Medicine
Background: Hemorrhagic myocardial infarction (hMI) can rapidly diminish the benefits of reperfusion therapy and direct the heart toward chronic heart failure. T2∗ cardiac magnetic resonance (CMR) is the reference standard for detecting hMI. However, the lack of clarity around the earliest time point for detection, time-dependent changes in hemorrhage volume, and the optimal methods for detection can limit the development of strategies to manage hMI.
Objectives: The authors investigated CMR signal characteristics of hMI through time-lapse multiparametric mapping using a clinically relevant animal model and evaluated the translatability in ST-segment elevation MI patients when possible.
Methods: Canines (N = 20) underwent 3.0-T CMR at baseline and various time points over the first week of reperfused MI. Time-dependent relationships between T1, T2, and T2∗ mapping of hMI, non-hMI, and remote territories were determined. Reperfused ST-segment elevation MI patients (N = 50) were studied to establish clinically feasibility.
Results: Although hMI was evident <1 hour after reperfusion on histopathology, it was not reliably detected with T1, T2, or T2∗ CMR. However, 24 hours to 7 days postreperfusion, hMI was detectable on T2∗ (27.0 ± 2.4 ms [baseline] vs 11.7 ± 2.8 ms [hMI]; P < 0.001), with stable volume and transmurality. In T2 maps, hMI was most visible 5 to 7 days postreperfusion with an area under the curve of 0.98 (sensitivity and specificity ≥0.95) relative to T2∗. However, this was not the case with T1 (sensitivity <0.8, across all time points).
Conclusions: HMI cannot be reliably detected with T1, T2, or T2∗ on 3.0-T CMR immediately after reperfusion. However, T2∗ CMR can be used to diagnose hMI between 24 hours and 7 days postreperfusion. T2 maps at 3.0-T are a strong alternative to T2∗ maps for diagnosing hMI, provided CMR is performed 5 to 7 days postreperfusion. However, diagnosing hMI with T1 is significantly more challenging at 3.0-T compared with both T2∗ and T2.
A pilot randomized controlled trial to explore the feasibility of a peer-delivered single-session brief intervention for youth with moderate risk substance use
(Public Library of Science, 2026-03-16) Jaguga, Florence; Turissini, Matthew; Kwobah, Edith Kamaru; Apondi, Edith; Enane, Leslie A.; Barasa, Julius; Kosgei, Gilliane; Olando, Yvonne; Ott, Mary A.; Kimaina, Allan; Aalsma, Matthew C.; Medicine, School of Medicine
Background: Youth in sub-Saharan Africa are at high risk of substance use yet lack access to appropriate interventions. The goal of this project was to evaluate the feasibility of a definitive trial to explore efficacy of a peer-delivered single-session brief intervention (SSBI) for youth with substance use in Kenya.
Methods: Seventy youth aged 15-24 years with moderate risk substance use were randomized to SSBI or to psychoeducation. Data was collected at baseline and month three. Primary outcomes: Feasibility criteria, e.g., study participation rate, proportion of participants willing to be randomized, and study completion rate. Strategies for recruitment in a future trial were collected using focus group discussions with the youth at month three. Secondary outcomes: (i) Change in substance use (Alcohol, Smoking & Substance Use Involvement Screening Test for Youth [ASSIST-Y] questionnaire), depression (Patient Health Questionnaire [PHQ-9]), anxiety (Generalized Anxiety Disorder [GAD-7 scale]), and quality of life (World Health Organization-Quality of Life Brief Version [WHO-QOL BREF]) scores between baseline and month 3; (ii) Fidelity to the intervention assessed using fidelity checklists.
Results: This pilot met most of the predefined minimum requirements for feasibility. For instance, 96.9% of those meeting eligibility criteria consented to participate (benchmark was 80%), and 100% of those who consented were willing to be randomized to either study arm. Youth reported that young people who use substances can be most effectively recruited from community settings. The SSBI showed a small effect on reducing total ASSIST-Y (Standardized Mean Difference [SMD] -0.33 95% Confidence Interval [CI] -0.83,0.16) scores in the intervention group compared to the control. There was a moderate improvement in the quality of life for the intervention group compared to the control (SMD -0.41 CI -0.91,0.09). The intervention had no effect on depression (SMD 0.23 CI -0.27,0.72) and anxiety symptoms (SMD 0.70 CI 0.19,1.2) at month 3.
Conclusion: It is feasible to conduct a randomized controlled trial of a peer-delivered SSBI for youth with moderate risk substance use in Kenya.