Baowei Fei

Medical hyperspectral imaging: a review

Abstract. Hyperspectral imaging (HSI) is an emerging imaging modality for medical applications, especially in disease diagnosis and image-guided surgery. HSI acquires a three-dimensional dataset called hypercube, with two spatial dimensions and one spectral dimension. Spatially resolved spectral imaging obtained by HSI provides diagnostic information about the tissue physiology, morphology, and composition. This review paper presents an overview of the literature on medical hyperspectral imaging technology and its applications. The aim of the survey is threefold: an introduction for those new to the field, an overview for those working in the field, and a reference for those searching for literature on a specific application.

Hyperspectral imaging and quantitative analysis for prostate cancer detection

Hyperspectral imaging (HSI) is an emerging modality for various medical applications. Its spectroscopic data might be able to be used to noninvasively detect cancer. Quantitative analysis is often necessary in order to differentiate healthy from diseased tissue. We propose the use of an advanced image processing and classification method in order to analyze hyperspectral image data for prostate cancer detection. The spectral signatures were extracted and evaluated in both cancerous and normal tissue. Least squares support vector machines were developed and evaluated for classifying hyperspectral data in order to enhance the detection of cancer tissue. This method was used to detect prostate cancer in tumor-bearing mice and on pathology slides. Spatially resolved images were created to highlight the differences of the reflectance properties of cancer versus those of normal tissue. Preliminary results with 11 mice showed that the sensitivity and specificity of the hyperspectral image classification method are 92.8% to 2.0% and 96.9% to 1.3%, respectively. Therefore, this imaging method may be able to help physicians to dissect malignant regions with a safe margin and to evaluate the tumor bed after resection. This pilot study may lead to advances in the optical diagnosis of prostate cancer using HSI technology.

Targeted iron-oxide nanoparticle for photodynamic therapy and imaging of head and neck cancer.

Photodynamic therapy (PDT) is a highly specific anticancer treatment modality for various cancers, particularly for recurrent cancers that no longer respond to conventional anticancer therapies. PDT has been under development for decades, but light-associated toxicity limits its clinical applications. To reduce the toxicity of PDT, we recently developed a targeted nanoparticle (NP) platform that combines a second-generation PDT drug, Pc 4, with a cancer targeting ligand, and iron oxide (IO) NPs. Carboxyl functionalized IO NPs were first conjugated with a fibronectin-mimetic peptide (Fmp), which binds integrin β1. Then the PDT drug Pc 4 was successfully encapsulated into the ligand-conjugated IO NPs to generate Fmp-IO-Pc 4. Our study indicated that both nontargeted IO-Pc 4 and targeted Fmp-IO-Pc 4 NPs accumulated in xenograft tumors with higher concentrations than nonformulated Pc 4. As expected, both IO-Pc 4 and Fmp-IO-Pc 4 reduced the size of HNSCC xenograft tumors more effectively than free Pc 4. Using a 10-fold lower dose of Pc 4 than that reported in the literature, the targeted Fmp-IO-Pc 4 NPs demonstrated significantly greater inhibition of tumor growth than nontargeted IO-Pc 4 NPs. These results suggest that the delivery of a PDT agent Pc 4 by IO NPs can enhance treatment efficacy and reduce PDT drug dose. The targeted IO-Pc 4 NPs have great potential to serve as both a magnetic resonance imaging (MRI) agent and PDT drug in the clinic.

MR/PET quantification tools: Registration, segmentation, classification, and MR-based attenuation correction

PURPOSE Combined MR∕PET is a relatively new, hybrid imaging modality. A human MR∕PET prototype system consisting of a Siemens 3T Trio MR and brain PET insert was installed and tested at our institution. Its present design does not offer measured attenuation correction (AC) using traditional transmission imaging. This study is the development of quantification tools including MR-based AC for quantification in combined MR∕PET for brain imaging. METHODS The developed quantification tools include image registration, segmentation, classification, and MR-based AC. These components were integrated into a single scheme for processing MR∕PET data. The segmentation method is multiscale and based on the Radon transform of brain MR images. It was developed to segment the skull on T1-weighted MR images. A modified fuzzy C-means classification scheme was developed to classify brain tissue into gray matter, white matter, and cerebrospinal fluid. Classified tissue is assigned an attenuation coefficient so that AC factors can be generated. PET emission data are then reconstructed using a three-dimensional ordered sets expectation maximization method with the MR-based AC map. Ten subjects had separate MR and PET scans. The PET with [(11)C]PIB was acquired using a high-resolution research tomography (HRRT) PET. MR-based AC was compared with transmission (TX)-based AC on the HRRT. Seventeen volumes of interest were drawn manually on each subject image to compare the PET activities between the MR-based and TX-based AC methods. RESULTS For skull segmentation, the overlap ratio between our segmented results and the ground truth is 85.2 ± 2.6%. Attenuation correction results from the ten subjects show that the difference between the MR and TX-based methods was <6.5%. CONCLUSIONS MR-based AC compared favorably with conventional transmission-based AC. Quantitative tools including registration, segmentation, classification, and MR-based AC have been developed for use in combined MR∕PET.

Spectral-spatial classification for noninvasive cancer detection using hyperspectral imaging

Abstract. Early detection of malignant lesions could improve both survival and quality of life of cancer patients. Hyperspectral imaging (HSI) has emerged as a powerful tool for noninvasive cancer detection and diagnosis, with the advantage of avoiding tissue biopsy and providing diagnostic signatures without the need of a contrast agent in real time. We developed a spectral-spatial classification method to distinguish cancer from normal tissue on hyperspectral images. We acquire hyperspectral reflectance images from 450 to 900 nm with a 2-nm increment from tumor-bearing mice. In our animal experiments, the HSI and classification method achieved a sensitivity of 93.7% and a specificity of 91.3%. The preliminary study demonstrated that HSI has the potential to be applied in vivo for noninvasive detection of tumors.

Characterization of the homogeneous tissue mixture approximation in breast imaging dosimetry

PURPOSE To compare the estimate of normalized glandular dose in mammography and breast CT imaging obtained using the actual glandular tissue distribution in the breast to that obtained using the homogeneous tissue mixture approximation. METHODS Twenty volumetric images of patient breasts were acquired with a dedicated breast CT prototype system and the voxels in the breast CT images were automatically classified into skin, adipose, and glandular tissue. The breasts in the classified images underwent simulated mechanical compression to mimic the conditions present during mammographic acquisition. The compressed thickness for each breast was set to that achieved during each patients last screening cranio-caudal (CC) acquisition. The volumetric glandular density of each breast was computed using both the compressed and uncompressed classified images, and additional images were created in which all voxels representing adipose and glandular tissue were replaced by a homogeneous mixture of these two tissues in a proportion corresponding to each breasts volumetric glandular density. All four breast images (compressed and uncompressed; heterogeneous and homogeneous tissue) were input into Monte Carlo simulations to estimate the normalized glandular dose during mammography (compressed breasts) and dedicated breast CT (uncompressed breasts). For the mammography simulations the x-ray spectra used was that used during each patients last screening CC acquisition. For the breast CT simulations, two x-ray spectra were used, corresponding to the x-ray spectra with the lowest and highest energies currently being used in dedicated breast CT prototype systems under clinical investigation. The resulting normalized glandular dose for the heterogeneous and homogeneous versions of each breast for each modality was compared. RESULTS For mammography, the normalized glandular dose based on the homogeneous tissue approximation was, on average, 27% higher than that estimated using the true heterogeneous glandular tissue distribution (Wilcoxon Signed Rank Test p = 0.00046). For dedicated breast CT, the overestimation of normalized glandular dose was, on average, 8% (49 kVp spectrum, p = 0.00045) and 4% (80 kVp spectrum, p = 0.000089). Only two cases in mammography and two cases in dedicated breast CT with a tube voltage of 49 kVp resulted in lower dose estimates for the homogeneous tissue approximation compared to the heterogeneous tissue distribution. CONCLUSIONS The normalized glandular dose based on the homogeneous tissue mixture approximation results in a significant overestimation of dose to the imaged breast. This overestimation impacts the use of dose estimates in absolute terms, such as for risk estimates, and may impact some comparative studies, such as when modalities or techniques with different x-ray energies are used. The error introduced by the homogeneous tissue mixture approximation in higher energy x-ray modalities, such as dedicated breast CT, although statistically significant, may not be of clinical concern. Further work is required to better characterize this overestimation and potentially develop new metrics or correction factors to better estimate the true glandular dose to breasts undergoing imaging with ionizing radiation.

3D ultrasound image segmentation using wavelet support vector machines.

PURPOSE Transrectal ultrasound (TRUS) imaging is clinically used in prostate biopsy and therapy. Segmentation of the prostate on TRUS images has many applications. In this study, a three-dimensional (3D) segmentation method for TRUS images of the prostate is presented for 3D ultrasound-guided biopsy. METHODS This segmentation method utilizes a statistical shape, texture information, and intensity profiles. A set of wavelet support vector machines (W-SVMs) is applied to the images at various subregions of the prostate. The W-SVMs are trained to adaptively capture the features of the ultrasound images in order to differentiate the prostate and nonprostate tissue. This method consists of a set of wavelet transforms for extraction of prostate texture features and a kernel-based support vector machine to classify the textures. The voxels around the surface of the prostate are labeled in sagittal, coronal, and transverse planes. The weight functions are defined for each labeled voxel on each plane and on the model at each region. In the 3D segmentation procedure, the intensity profiles around the boundary between the tentatively labeled prostate and nonprostate tissue are compared to the prostate model. Consequently, the surfaces are modified based on the model intensity profiles. The segmented prostate is updated and compared to the shape model. These two steps are repeated until they converge. Manual segmentation of the prostate serves as the gold standard and a variety of methods are used to evaluate the performance of the segmentation method. RESULTS The results from 40 TRUS image volumes of 20 patients show that the Dice overlap ratio is 90.3% ± 2.3% and that the sensitivity is 87.7% ± 4.9%. CONCLUSIONS The proposed method provides a useful tool in our 3D ultrasound image-guided prostate biopsy and can also be applied to other applications in the prostate.

An MR image-guided, voxel-based partial volume correction method for PET images

PURPOSE Partial volume effect in positron emission tomography (PET) can cause incorrect quantification of radiopharmaceutical uptake in functional imaging. A PET partial volume correction method is presented to attenuate partial volume blurring and to yield voxel-based corrected PET images. METHODS By modeling partial volume effect as a convolution of point spread function of the PET scanner, the reconstructed PET images are corrected by iterative deconvolution with an edge-preserving smoothness constraint. The constraint is constructed to restore discontinuities extracted from coregistered MR images but maintains the smoothness in radioactivity distribution. The correction is implemented in a Bayesian deconvolution framework and is solved by a conjugate gradient method. The performance of the method was compared with the geometric transfer matrix (GTM) method on a simulated dataset. The method was evaluated on synthesized brain FDG-PET data and phantom MRI-PET experiments. RESULTS The true PET activity of objects with a size of greater than the full-width at half maximum of the point spread function has been effectively restored in the simulated data. The partial volume correction method is quantitatively comparable to the GTM method. For synthesized FDG-PET with true activity 0 μci/cc for cerebrospinal fluid (CSF), 228 μci/cc for white matter (WM), and 621 μci/cc for gray matter (GM), the method has improved the radioactivity quantification from 186 ± 16 μci/cc to 30 ± 7 μci/cc in CSF, 317 ± 15 μci/cc to 236 ± 10 μci/cc for WM, 438 ± 4 μci/cc to 592 ± 5 μci/cc for GM. Both visual and quantitative assessments show improvement of partial volume correction in the synthesized and phantom experiments. CONCLUSIONS The partial volume correction method improves the quantification of PET images. The method is comparable to the GTM method but does not need MR image segmentation or prior tracer distribution information. The voxel-based method can be particularly useful for combined PET/MRI studies.

Superpixel-Based Segmentation for 3D Prostate MR Images

This paper proposes a method for segmenting the prostate on magnetic resonance (MR) images. A superpixel-based 3D graph cut algorithm is proposed to obtain the prostate surface. Instead of pixels, superpixels are considered as the basic processing units to construct a 3D superpixel-based graph. The superpixels are labeled as the prostate or background by minimizing an energy function using graph cut based on the 3D superpixel-based graph. To construct the energy function, we proposed a superpixel-based shape data term, an appearance data term, and two superpixel-based smoothness terms. The proposed superpixel-based terms provide the effectiveness and robustness for the segmentation of the prostate. The segmentation result of graph cuts is used as an initialization of a 3D active contour model to overcome the drawback of the graph cut. The result of 3D active contour model is then used to update the shape model and appearance model of the graph cut. Iterations of the 3D graph cut and 3D active contour model have the ability to jump out of local minima and obtain a smooth prostate surface. On our 43 MR volumes, the proposed method yields a mean Dice ratio of 89.3 ±1.9%. On PROMISE12 test data set, our method was ranked at the second place; the mean Dice ratio and standard deviation is 87.0±3.2%. The experimental results show that the proposed method outperforms several state-of-the-art prostate MRI segmentation methods.

Detection of Cancer Metastasis Using a Novel Macroscopic Hyperspectral Method

The proposed macroscopic optical histopathology includes a broad-band light source which is selected to illuminate the tissue glass slide of suspicious pathology, and a hyperspectral camera that captures all wavelength bands from 450 to 950 nm. The system has been trained to classify each histologic slide based on predetermined pathology with light having a wavelength within a predetermined range of wavelengths. This technology is able to capture both the spatial and spectral data of tissue. Highly metastatic human head and neck cancer cells were transplanted to nude mice. After 2- 3 weeks, the mice were euthanized and the lymph nodes and lung tissues were sent to pathology. The metastatic cancer is studied in lymph nodes and lungs. The pathological slides were imaged using the hyperspectral camera. The results of the proposed method were compared to the pathologic report. Using hyperspectral images, a library of spectral signatures for different tissues was created. The high-dimensional data were classified using a support vector machine (SVM). The spectra are extracted in cancerous and non-cancerous tissues in lymph nodes and lung tissues. The spectral dimension is used as the input of SVM. Twelve glasses are employed for training and evaluation. The leave-one-out cross-validation method is used in the study. After training, the proposed SVM method can detect the metastatic cancer in lung histologic slides with the specificity of 97.7% and the sensitivity of 92.6%, and in lymph node slides with the specificity of 98.3% and the sensitivity of 96.2%. This method may be able to help pathologists to evaluate many histologic slides in a short time.