Photoacoustic (PA) imaging is a rapidly emerging non-invasive soft tissue imaging modality which has the potential to detect tissue abnormality at early stage. Photoacoustic images map the spatially varying optical absorption property of tissue. In multiwavelength photoacoustic imaging, the soft tissue is imaged with different wavelengths, tuned to the absorption peaks of the specific light absorbing tissue constituents or chromophores to obtain images with different contrasts of the same tissue sample. From those images, spatially varying concentration of the chromophores can be recovered. As multiwavelength PA images can provide important physiological information related to function and molecular composition of the tissue, so they can be used for diagnosis of cancer lesions and differentiation of malignant tumors from benign tumors.
In this research, a number of parameters have been extracted from multiwavelength 3D PA images of freshly excised human prostate and thyroid specimens, imaged at five different wavelengths. Using marked histology slides as ground truths, region of interests (ROI) corresponding to cancer, benign and normal regions have been identified in the PA images. The extracted parameters belong to different categories namely chromophore concentration, frequency parameters and PA image pixels and they represent different physiological and optical properties of the tissue specimens. Statistical analysis has been performed to test whether the extracted parameters are significantly different between cancer, benign and normal regions. A multidimensional [29 dimensional] feature set, built with the extracted parameters from the 3D PA images, has been divided randomly into training and testing sets. The training set has been used to train support vector machine (SVM) and neural network (NN) classifiers while the performance of the classifiers in differentiating different tissue pathologies have been determined by the testing dataset. Using the NN classifier, performance of parameters belonging to different categories in differentiating malignant tissue from nonmalignant tissue has been determined. It has been found that, among different categories, the frequency parameters performed best in differentiating malignant from nonmalignant tissue [sensitivity and specificity with testing dataset are 85% and 84%] while performance of all the categories combined was better than that [sensitivity and specificity with testing dataset are 93% and 91%].
However, PA imaging cannot be used to provide the anatomical cues required to determine the position of the detected or suspected malignant tumor region relative to familiar organ landmarks. On the other hand, although accuracy of Ultrasound (US) imaging in detecting cancer lesions is low, major anatomical cues like organ boundaries or presence of nearby major organs are visible in US images. A dual mode PA and US imaging system can potentially detect as well as localize cancer lesions with high accuracy. In this study, we have developed a novel pulse echo US imaging system which can be easily integrated with our existing ex-vivo PA imaging system to produce the dual mode imaging system. Here a Polyvinylidene fluoride (PVDF) film has been used as US transmitter. To improve the anticipated low signal to noise ratio (SNR) of the received US signal due to the low electromechanical coupling coefficient of the PVDF film, we implemented pulse compression technique using chirp signals. Comparisons among the different SNR values obtained with short pulse and after pulse compression with chirp signal show a clear improvement of the SNR for the compressed pulse. The axial resolution of the imaging system improved with increasing sweep bandwidth of input chirp signals, whereas the lateral resolution remained almost constant. This work demonstrates the feasibility of using a PVDF film transducer as an US transmitter and implementing pulse compression technique in an acoustic lens focusing based imaging system.
Library of Congress Subject Headings
Diagnostic imaging; Optoacoustic spectroscopy; Tissues--Imaging; Cancer--Diagnosis
Imaging Science (Ph.D.)
Department, Program, or Center
Chester F. Carlson Center for Imaging Science (COS)
S. Manian Ramkumar
Sinha, Saugata, "Photoacoustic Image Analysis for Cancer Detection and Building a Novel Ultrasound Imaging System" (2014). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus