Skip to main content

Staff Profile

Dr Anando Sen

Data Scientist

  • Address: Translational and Clinical Research Institute
    John Walton Muscular Dystrophy Research Centre
    Faculty of Medical Sciences
    Newcastle University
    International Centre for Life
    NE1 3BZ
    United Kingdom
Background
Current Role

Data Scientist for the Conect4children project (John Walton Muscular Dystrophy Research Center)

Conect4children (c4c) is an Innovative Medicines Initiative funded project that is addressing the barriers to conducting paediatric research by creating a pan-European clinical trial network. Newcastle University is tasked with improving the standardisation and harmonisation of data collected in paediatric clinical trials. This includes:
  • Implementing the FAIR data principles to increase the ufility of data (for both humans and computers) by addressing the Findability, Accessibility, Interoperability and Reusability of the data.
  • Building a Cross Cutting Paediatric Data Dictionary to be used in proof of viability paediatric trials.
  • Developing a new global set of standards for paediatric data in collaboration with CDISC (Clinical Data Interchange Standards Consortium).
  • Exploring ways to use tools, ontologies and better semantic interoperability to standardise disease specfic data.
  • Research and implement solutions to improving linkage between real world data (especially electronic health records) and the data collected in clinical trials.
These tasks should result in data that is more standardised, and interoperable. This potentially poolable data can then be used (and re-used) to allow for better planning of clinical trials and increased knowledge about paediatric medicines development. This work feeds into the goal of c4c: “Better medicines for babies, children and young people”


Background

I completed my PhD in Mathematics in 2012 from University of Houston. I have since held postdoctoral positions at University of Houston (Biomedica Engineering) and Columbia University (Biomedical Informatics). Most recently I was a Research Investigator (Imaging Physics) at MD Anderson Cancer Center.

Research

My research involves developing mathematical models for various applications in healthcare - particularly medical imaging and clinical trials. The models are implemented in fifth generation programming languages and are accompanied by rigorous validations, statistical analysis and error reporting. My projects have included:

  • Standardization and harmonization of paediatric clinical data - Please see under current role.
  • Radiology-pathology correlation for tumor delineations - Despite tremendous advancement of in vivo imaging modalities, there remains substantial uncertainty in tumor delineations in these images. Histopathology is the gold-standard for determining the extent of cancerous tumors. Correlating in vivo imaging with histopathology has the potential to provide a direct spatial comparison between the two. We use clinical and pre-clinical data to develop and validate correlative pathology methods.
  • Deformable image registration for liver cancer radiation therapy - Radiation therapy is often used to treat inoperable cholangiocarcinomas (bile-duct cancer). Due to the vast changes the liver can undergo during irradiation, deformable image registration an attractive option for comparing pre and post-treatment images, hence assessing the treatment. We develop and compare deformable registration techniques for this purpose.
  • Quantitative assessment of population representativeness in individual clinical trials - An apriori estimate for the population representativeness of a clinical trial can be a major aid in patient selection strategies. We develop a metric GIST 2.0, to compute the population representativeness based on eligibility criteria. The iterative use of this metric during the design of eligibility criteria can potentially optimize population representativeness within the constraints of patient safety.
  • Tumor localization tasks in prostate nuclear medicine images - Model Observers are mathematical models used for tumor localization tasks in medical imaging. Since medical images are analyzed by radiologists, it is crucial for model observers to be relevant for humans. We develop a visual-search observer and implement inefficiency models within it to account for human perceptions.
  • CT reconstructions with truncated projection data - CT scans have tremendous diagnostic value in the detection of tumors but are accompanied by risks of radiation-induced carcinomas. Restricting CT scans to a small region of interest can subtantially reduce the exposure to irradiation. We present a new algorithm to reconstruct anatomical images from such restricted scans that can maintain image quality comparable to full CT scans. The performance of the algorithm and its robustness to noise are validated on both synthetic and experimental data.
Publications