• Devalla Hitesh Sri Sai
• Chirag M
• Anoop Soudri
• Abhimanyu Hiremath
• Ms. Dhivya V

Deep learning models are proficient at segmenting brain tumors; however, they face challenges in clinical environments due to the "black box" problem and a lack of adequate testing involving real users. This research investigates a pseudocolor-guided U-Net segmentation approach by assessing its technical effectiveness and interpretability for users. The model, which was trained on the BraTS 2021 dataset and utilized confidence-based post-processing, recorded Dice scores of 0.87 for the necrotic core, 0.82 for edema, and 0.85 for the enhancing tumor. A study conducted with five neuroradiologists demonstrated that pseudocolor visualization enhanced boundary identification accuracy by 23% (p<0.001). Additionally, it reduced diagnostic time by 29% (p<0.01) and increased confidence scores from 3.1 out of 5 to 4.5 out of 5 (p<0.001). Interrater agreement improved from κ=0.68 to κ=0.84. These results indicate that pseudocolor visualization significantly enhances comprehension and decision-making. This emphasizes the importance of validating AI technologies in medical imaging with an emphasis on human usability.

Brain tumor segmentation; Explainable AI; Pseudocolor visualization; User study validation; Deep learning interpretability; Medical imaging; Neuroradiology

"Pseudocolor - Assisted Segmentation of Brain Tumors, Validation of Interpretability, and Examination of Radiologist Decision-Making.", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.12, Issue 12, page no.d46-d50, December-2025, Available :http://www.jetir.org/papers/JETIR2512308.pdf

"Pseudocolor - Assisted Segmentation of Brain Tumors, Validation of Interpretability, and Examination of Radiologist Decision-Making.", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.12, Issue 12, page no. ppd46-d50, December-2025, Available at : http://www.jetir.org/papers/JETIR2512308.pdf