Date of Award

Winter 12-15-2021

Author's School

McKelvey School of Engineering

Author's Department

Biomedical Engineering

Degree Name

Doctor of Philosophy (PhD)

Degree Type



Laser stimulated dynamic thermal imaging system for tumor detectionby Hongyu Meng Doctor of Philosophy in Biomedical Engineering Washington University in St. Louis, 2021 Professor Samuel Achilefu, Chair Recent advances in infrared sensor technology have enabled the rapid application of thermal imaging in materials science, security and medicine. Relying on the infrared characteristics of living systems, thermal imaging has been used to generate individual heat maps, detect inflammation and tumor. As an imaging system, thermal imaging has the advantages of portability, real-time, non-invasive, and non-contact. But the low specificity of thermal imaging hinders its wide clinical application.

Unfortunately, label-free DTI is less able to fully capture thermal tissue heterogeneity in high resolution due partly to how thermal stimulation is applied. Current DTI methods apply thermal stimulation to a large area of tissue, which obscures the detection of the unique thermal characteristics of a small area in the thermally disturbed area. Super-resolution DTI grating can improve the spatial resolution, but the system setup is complex. For biological samples, the use of exogenous contrast agents can enhance contrast, but contrast agents increase regulatory hurdles in clinical trials.

In this work, we have developed a focused dynamic laser stimulation imaging (FDTI) system to overcome these limitations. The system, which has high resolution, high speed and large field of view uses a short wavelength laser to stimulate small tissue area and a thermal camera to acquire data. We captured thermal images and videos, extracted features, and built classifiers to distinguish tumors from normal tissues. Data analysis showed that FDTI method achieved high accuracy (classifier surpassed 90%) with spatial resolution attaining 1 mm, which surpasses conventional thermal imaging and DTI.

We next explored the ability of FDTI to detect early-stage tumors by scanning multiple areas that exhibited normal thermal images with conventional thermal imaging. A bioluminescence imaging (BLI) system was then used to locate the tumor, which was co-registered to the FDTI images to determine the position of the laser spot. By extracting features from the collected thermal images and videos and constructing the classifier, the FDTI system achieved an accuracy greater than 80% in detecting early tumors in different mouse tumor models.

Subsequently, the FDTI system was optimized to improve its acquisition speed, automation and robustness. First, we analyzed the influencing factors of imaging and proposed new system hardware designs to improve the data acquisition speed. Then, to shorten the acquisition time from the software level, we tested and analyzed the performance of features at different stages during the acquisition process. We also designed and tested registration markers, including registration results of different features, feature robustness under interference, marker detection from the background, and marker performance in motion correction to improve the degree of automation of the system. Furthermore, we tested the performance of thermal imaging applications in other research fields, including brain tumor detection, nerve damage assessment, and whether temperature changes correlate with stroke.

These results show that FDTI is a promising technique for enhancing contrast, improving spatial resolution, determining underlying tumor heterogeneity, and detecting tumors at stages when conventional thermal imaging is ineffective. This work lays a strong foundation for diverse applications and clinical translation of FDTI to address unmet needs of current thermal imaging technologies.


English (en)


Samuel Achilefu

Committee Members

Abhinav Kumar Jha

Included in

Engineering Commons