Mina Heinein

M. Heinein*, M. Youssef*, K. Nashed*, T. Basha, H. M. T. Alam, A. M. Selim, O. S. Bhatti, D. Sonntag

• MICCAI 2026, MSB EMERGE Workshop — ORAL  *Equal contribution. First author. Graduation thesis, in collaboration with DFKI; funded by ASRT & ITIDA.
• Patient-specific multimodal hypergraphs for medical visual question answering, unifying structured EHR events, prior-report entities, cross-modal grounding, SNOMED CT ontology concepts, and image regions in one temporally indexed structure.
• Temporally-aware multimodal retrieval with leakage control, scoring evidence by textual, visual, temporal, and hypergraph proximity to build compact evidence subgraphs for VLM answer generation.
• 73.95% accuracy on the EHRXQA benchmark (95% CI 71.93–75.87), beating an RL-trained state of the art (70.43%) without task-specific training; +10.0 open-ended weighted-F1 at 27B, and multi-study F1 from 5.2 to 43.4.
• Paper · Code · Demo video

K. Nashed*, M. Heinein*, M. Youssef*, T. Basha, H. M. T. Alam, A. M. Selim, O. S. Bhatti, D. Sonntag

• MICCAI 2026, MSB EMERGE Workshop — ORAL  *Equal contribution. Second author, with DFKI.
• Adapts SAM3's concept grounding using only bounding boxes and concept names — no pixel masks. The mask decoder and over 95% of the vision backbone stay frozen.
• On a 10-class MIMIC-CXR benchmark, text-only mIoU improves over Medical-SAM3 from 0.48 to 0.69 for anatomical structures and from 0.09 to 0.53 for pathological findings.
• Paper · Code & supplementary

Paris, France

Cairo, Egypt

• Relevant coursework: Machine Learning, Deep Learning, Computer Vision, Medical Imaging, Data Structures, Algorithms, Object-Oriented Programming, Digital Signal Processing, Embedded Systems, Computer Architecture, Biomedical Instrumentation.

• Built CRM solutions on Microsoft Dynamics 365 with retail and corporate customer portals, implementing field specifications, form layouts, and validation rules for contact and account management.
• Developed REST and SOAP web services for case data retrieval, created web resources for dashboard integration, and implemented custom .NET plugins and automated business rule enforcement.
• Created an end-to-end case management workflow with automated routing rules, status tracking, SLA implementation, and email notifications, supporting multiple case types including credit card operations.

• Developed dynamic web applications, integrating assistive technology for enhanced accessibility.
• Worked across web development, IoT, and assistive technology, focused on users with mobility, sensory, or cognitive challenges.

• Multi-port signal viewer: real-time ECG/EEG analyzer with automated reporting.
• Signal equalizer: vocal extraction, animal/music sound separation, Wiener noise filtering, real-time frequency manipulation, dynamic spectrograms.
• Shazam-style audio fingerprinting using similarity matching for music identification.
• Repository — built with a team of five.

• Language learning through song lyrics. Integrated the Spotify API for login, playlist access, and playback, increasing user engagement by 45%.
• Scored singing accuracy in real time using Google Cloud Speech and Levenshtein distance; clickable lyrics give instant dictionary definitions.
• Repository

• Full-stack e-commerce platform. Custom orders let users personalise products, increasing conversion 25%; the admin dashboard cut order-processing time 60%.
• Repository

• Bare-metal driver development for LCD, DIO, interrupts, and UART; ESP32 for Wi-Fi, enabling remote monitoring and control from a web interface.
• Repository

• Classical computer vision from first principles: histogram equalisation, spatial and frequency filtering, thresholding, Sobel/Canny/active-contour edge detection, Harris and SIFT detection and matching, K-means/region-growing/Otsu segmentation, and a face recognition pipeline.

• Complete rules engine — legal move generation, check and checkmate detection, castling, en-passant, promotion — with a Swing GUI, move history, and per-player timers.
• Repository

• Twelve-month project-based curriculum across programming languages, data structures, and algorithms; systems engineering and DevOps, Nginx configuration, and load balancing.

• Microcontroller interfacing and real-time operating systems; coursework in C programming, embedded system architecture, and automotive bus technology.