Tamjid Imtiaz

Non-hematopoietic cells are essential contributors to hematopoiesis. However, heterogeneity and spatial organization of these cells in human bone marrow remain largely uncharacterized. We used single-cell RNA sequencing (scRNA-seq) to profile 29,325 non-hematopoietic cells and discovered nine transcriptionally distinct subtypes. We simultaneously profiled 53,417 hematopoietic cells and predicted their interactions with non-hematopoietic subsets. We employed co-detection by indexing (CODEX) to spatially profile over 1.2 million cells. We integrated scRNA-seq and CODEX data to link predicted cellular signaling with spatial proximity. Our analysis revealed a hyperoxygenated arterio-endosteal neighborhood for early myelopoiesis, and an adipocytic localization for early hematopoietic stem and progenitor cells (HSPCs). We used our CODEX atlas to annotate new images and uncovered mesenchymal stromal cell (MSC) expansion and spatial neighborhoods co-enriched for leukemic blasts and MSCs in acute myeloid leukemia (AML) patient samples. This spatially resolved, multiomic atlas of human bone marrow provides a reference for investigation of cellular interactions that drive hematopoiesis.

Brain tumor segmentation from Magnetic Resonance Imaging (MRI) is of great importance for better tumor diagnosis, growth rate prediction and radiotherapy planning. But this task is extremely challenging due to intrinsically heterogeneous tumor appearance, the presence of severe partial volume effect and ambiguous tumor boundaries. In this work, a unique approach of tumor segmentation is introduced based on superpixel level features extracted from all three planes (x -y, y - z, and z - x) of 3D volumetric MR images. In order to avoid the pixel randomness and to account for precise inhomogeneous boundaries of brain tumor, each of the images belonging to a particular plane is partitioned into irregular patches (superpixels) based on their intensity and spatial similarity. Next, various statistical and textural features are extracted from each superpixel where all three planes are considered separately in order to obtain better labeling on superpixels in tumor edges. A feature selection scheme is proposed based on their performance on histogram based consistency analysis and local descriptor pattern analysis, which offers a significant reduction in feature dimension without sacrificing classification performance. For the purpose of supervised classification, Extremely Randomized Trees is used to classify these superpixels into a tumor or a non-tumor class. Finally, pixel level decision is taken based on corresponding decisions obtained in each plane. Extensive simulations are carried out on publicly available dataset and it is found that the proposed method offers better tumor segmentation performance in comparison to that obtained by some state of the art methods.

Since the end of 2019, novel coronavirus disease (COVID-19) has brought about a plethora of unforeseen changes to the world as we know it. Despite our ceaseless fight against it, COVID-19 has claimed millions of lives, and the death toll exacerbated due to its extremely contagious and fast-spreading nature. To control the spread of this highly contagious disease, a rapid and accurate diagnosis can play a very crucial part. Motivated by this context, a parallelly concatenated convolutional block-based capsule network is proposed in this article as an efficient tool to diagnose the COVID-19 patients from multimodal medical images. Concatenation of deep convolutional blocks of different filter sizes allows us to integrate discriminative spatial features by simultaneously changing the receptive field and enhances the scalability of the model. Moreover, concatenation of capsule layers strengthens the model to learn more complex representation by presenting the information in a fine to coarser manner. The proposed model is evaluated on three benchmark datasets, in which two of them are chest radiograph datasets and the rest is an ultrasound imaging dataset. The architecture that we have proposed through extensive analysis and reasoning achieved outstanding performance in COVID-19 detection task, which signifies the potentiality of the proposed model.

Tuberculosis (TB) is a communicable disease that is one of the top 10 causes of death worldwide according to the World Health Organization. Hence, Early detection of Tuberculosis is an important task to save millions of lives from this life threatening disease. For diagnosing TB from chest X-Ray, different handcrafted features were utilized previously and they provided high accuracy even in a small dataset. However, at present, deep learning (DL) gains popularity in many computer vision tasks because of their better performance in comparison to the traditional manual feature extraction based machine learning approaches and Tuberculosis detection task is not an exception. Considering all these facts, a cascaded ensembling method is proposed that combines both the hand-engineered and the deep learning-based features for the Tuberculosis detection task. To make the proposed model more generalized, rotation-invariant augmentation techniques are introduced which is found very effective in this task. By using the proposed method, outstanding performance is achieved through extensive simulation on two benchmark datasets (99.7% and 98.4% accuracy on Shenzhen and Montgomery County datasets respectively) that verifies the effectiveness of the method.