As a result of the expansive information set size of high-resolution image data, most desktop workstations do not have sufficient configurable scheduling to perform image processing assignments in a convenient manner due to which the image processing tasks are meant to be divided into straight forward assignments. The processing power of any regular computing machine in this way becomes a severe bottleneck with respect to high execution time and low throughput. Many image processing tasks exhibit a high level of information region and parallelism and map quite readily to a parallel computing system. This paper shows an alternative to sequential image processing by introducing Map-Reduce technique to segment multiple images with the help of Hadoop framework. The evaluation of the proposed scheduling algorithm is done by implementing parallel image segmentation algorithm to detect lung tumor for up to 1 GB size of CT image dataset. The results have shown improved performance with parallel image segmentation when compared to sequential image segmentation method particularly when data capacity reaches a particular threshold. This is because the process of parallel image processing has been able to exploit the multi-cores thread level parallelism which ultimately gave the CPU usage with octacores up to 96%, hence reducing the task execution time up to approximately 1.6 times compared with the sequential style of image segmentation using Map-Reduce algorithm implemented with FIFO scheduler. The proposed parallel image segmentation design has shown to be useful for researchers at performing bulk image segmentation in parallel, which can save tremendous execution time. © 2019, North Atlantic University Union. All right reserved.