Breadcrumb

Jehangir Khan: Automated dosimetry makes radiation therapy more personalized

Published

Jehangir Khan has developed methods that make it faster and more reliable to calculate the radiation dose to the kidneys during treatment of neuroendocrine tumors. The findings may contribute to more personalized and safer care.

Image
Jehangir Khan, doctoral student at the Institute of Clinical Sciences and medical physicist at Örebro University Hospital.

JEHANGIR KHAN
Dissertation defense: 17 September 2025 (click for details)
Doctoral thesis: Image-based kidney dosimetry method for  [¹⁷⁷Lu]Lu-DOTATATE treatments Development, validation, and assessment of quantification accuracy and precision
Research area: Medical Radiation Sciences
Sahlgrenska Academy, The Institute of Clinical Sciences

Neuroendocrine tumors (NETs) are a rare group of cancers often diagnosed at a late stage. In Sweden, about 400–500 new cases are discovered each year. Because tumor cells often have many receptors on their surface for the hormone somatostatin (somatostatin receptors), patients can be treated with radionuclide therapy. This involves attaching a radioactive isotope to a substance that binds specifically to the tumor. One example is the drug [¹⁷⁷Lu]Lu-DOTATATE, which was approved in both Europe and the United States in 2018.

“With dosimetry, meaning the calculation of how much radiation different organs absorb, we can optimize treatment by increasing the dose to the tumor while reducing exposure to organs at risk, such as the kidneys and bone marrow,” says Jehangir Khan, medical physicist at Örebro University Hospital and doctoral student at the Institute of Clinical Sciences.

Figure from the thesis. Illustration of how a 3D convolutional neural network is used to automatically segment the kidneys in SPECT/CT images.

Faster and more accurate segmentation

One barrier to dosimetry in clinical practice today is that the method is time-consuming. It requires multiple imaging sessions, analyses, and manual calculations. That is why Jehangir Khan and his colleagues developed a method to automate kidney segmentation in CT images.

“We used a convolutional neural network (CNN), a form of artificial intelligence for image analysis. It makes segmentation both faster and more consistent than when done manually, and it streamlines the entire dosimetry process for treatment with [¹⁷⁷Lu]Lu-DOTATATE.”

Figure from the thesis. An overview diagram illustrating the fundamental steps in the dosimetry workflow.

Optimized method for dose calculation

The research group also refined a method where small volumes are delineated in the kidney tissue (parenchyma). By measuring how the drug distributes in these small volumes, it is possible to calculate how much radiation the kidneys are exposed to.

“We optimized this so-called small-VOI method so that it provides results as reliable as the more labor-intensive method where the entire kidney tissue is delineated (the WKP method). In addition, we developed an empirical model to estimate uncertainty in the calculations at the individual level. That is crucial for optimizing treatment and improving patient safety,” says Jehangir Khan.

Combining clinic and research

What has been rewarding and what has been challenging about your doctoral project?
“It has been very rewarding to combine clinical work with research. Developing a method that is both scientifically grounded and practically useful in daily work has been especially motivating. At the same time, it has been challenging to find time for research, particularly during periods of heavy workload, but it has also been highly educational.”

Text: Jakob Lundberg