Oral mucosal lesions triage via YOLOv7 models.

BACKGROUND/PURPOSE: The global incidence of lip and oral cavity cancer continues to rise, necessitating improved early detection methods. This study leverages the capabilities of computer vision and deep learning to enhance the early detection and classification of oral mucosal lesions. METHODS: A dataset initially consisting of 6903 white-light macroscopic images collected from 2006 to 2013 was expanded to over 50,000 images to train the YOLOv7 deep learning model. Lesions were categorized into three referral grades: benign (green), potentially malignant (yellow), and malignant (red), facilitating efficient triage. RESULTS: The YOLOv7 models, particularly the YOLOv7-E6, demonstrated high precision and recall across all lesion categories. The YOLOv7-D6 model excelled at identifying malignant lesions with notable precision, recall, and F1 scores. Enhancements, including the integration of coordinate attention in the YOLOv7-D6-CA model, significantly improved the accuracy of lesion classification. CONCLUSION: The study underscores the robust comparison of various YOLOv7 model configurations in the classification to triage oral lesions. The overall results highlight the potential of deep learning models to contribute to the early detection of oral cancers, offering valuable tools for both clinical settings and remote screening applications.

Online geometry calibration for retrofit computed tomography from a mouse rotation system and a small-animal imager.

BACKGROUND: Computed tomography (CT) generates a three-dimensional rendering that can be used to interrogate a given region or desired structure from any orientation. However, in preclinical research, its deployment remains limited due to relatively high upfront costs. Existing integrated imaging systems that provide merged planar X-ray also dwarfs CT popularity in small laboratories due to their added versatility. PURPOSE: In this paper, we sought to generate CT-like data using an existing small-animal X-ray imager with a specialized specimen rotation system, or MiSpinner. This setup conforms to the cone-beam CT (CBCT) geometry, which demands high spatial calibration accuracy. Therefore, a simple but robust geometry calibration algorithm is necessary to ensure that the entire imaging system works properly and accurately. METHODS: Because the rotation system is not permanently affixed, we propose a structure tensor-based two-step online (ST-TSO) geometry calibration algorithm. Specifically, two datasets are needed, namely, calibration and actual measurements. A calibration measurement detects the background of the system forward X-ray projections. A study on the background image reveals the characteristics of the X-ray photon distribution, and thus, provides a reliable estimate of the imaging geometry origin. Actual measurements consisted of an X-ray of the intended object, including possible geometry errors. A comprehensive image processing technique helps to detect spatial misalignment information. Accordingly, the first processing step employs a modified projection matrix-based calibration algorithm to estimate the relevant geometric parameters. Predicted parameters are then fine-tuned in a second processing step by an iterative strategy based on the symmetry property of the sum of projections. Virtual projections calculated from the parameters after two-step processing compensate for the scanning errors and are used for CT reconstruction. Experiments on phantom and mouse imaging data were performed to validate the calibration algorithm. RESULTS: Once system correction was conducted, CBCT of a CT bar phantom and a cohort of euthanized mice were analyzed. No obvious structure error or spatial artifacts were observed, validating the accuracy of the proposed geometry calibration method. Digital phantom simulation indicated that compared with the preset spatial values, errors in the final estimated parameters could be reduced to 0.05° difference in dominant angle and 0.5-pixel difference in dominant axis bias. The in-plane resolution view of the CT-bar phantom revealed that the resolution approaches 150 µ $\umu$ m. CONCLUSIONS: A constrained two-step online geometry calibration algorithm has been developed to calibrate an integrated X-ray imaging system, defined by a first-step analytical estimation and a second-step iterative fine-tuning. Test results have validated its accuracy in system correction, thus demonstrating the potential of the described system to be modified and adapted for preclinical research.

Few-shot transfer learning for personalized atrial fibrillation detection using patient-based siamese network with single-lead ECG records.

The proliferation of wearable devices has allowed the collection of electrocardiogram (ECG) recordings daily to monitor heart rhythm and rate. For example, 24-hour Holter monitors, cardiac patches, and smartwatches are widely used for ECG gathering and application. An automatic atrial fibrillation (AF) detector is required for timely ECG interpretation. Deep learning models can accurately identify AFs if large amounts of annotated data are available for model training. However, it is impractical to request sufficient labels for ECG recordings for an individual patient to train a personalized model. We propose a Siamese-network-based approach for transfer learning to address this issue. A pre-trained Siamese convolutional neural network is created by comparing two labeled ECG segments from the same patient. We sampled 30-second ECG segments with a 50% overlapping window from the ECG recordings of patients in the MIT-BIH Atrial Fibrillation Database. Subsequently, we independently detected the occurrence of AF in each patient in the Long-Term AF Database. By fine-tuning the model with the 1, 3, 5, 7, 9, or 11 ECG segments ranging from 30 to 180 s, our method achieved macro-F1 scores of 96.84%, 96.91%, 96.97%, 97.02%, 97.05%, and 97.07%, respectively.

Direct mapping from PET coincidence data to proton-dose and positron activity using a deep learning approach.

Objective. Obtaining the intrinsic dose distributions in particle therapy is a challenging problem that needs to be addressed by imaging algorithms to take advantage of secondary particle detectors. In this work, we investigate the utility of deep learning methods for achieving direct mapping from detector data to the intrinsic dose distribution.Approach. We performed Monte Carlo simulations using GATE/Geant4 10.4 simulation toolkits to generate a dataset using human CT phantom irradiated with high-energy protons and imaged with compact in-beam PET for realistic beam delivery in a single-fraction (∼2 Gy). We developed a neural network model based on conditional generative adversarial networks to generate dose maps conditioned on coincidence distributions in the detector. The model performance is evaluated by the mean relative error, absolute dose fraction difference, and shift in Bragg peak position.Main results. The relative deviation in the dose and range of the distributions predicted by the model from the true values for mono-energetic irradiation between 50 and 122 MeV lie within 1% and 2%, respectively. This was achieved using 105coincidences acquired five minutes after irradiation. The relative deviation in the dose and range for spread-out Bragg peak distributions were within 1% and 2.6% uncertainties, respectively.Significance. An important aspect of this study is the demonstration of a method for direct mapping from detector counts to dose domain using the low count data of compact detectors suited for practical implementation in particle therapy. Including additional prior information in the future can further expand the scope of our model and also extend its application to other areas of medical imaging.

Image reconstruction in phase-contrast tomography exploiting the second-order statistical properties of the projection data.

X-ray phase-contrast tomography (PCT) methods seek to quantitatively reconstruct separate images that depict an object's absorption and refractive contrasts. Most PCT reconstruction algorithms generally operate by explicitly or implicitly performing the decoupling of the projected absorption and phase properties at each tomographic view angle by use of a phase-retrieval formula. However, the presence of zero-frequency singularity in the Fourier-based phase retrieval formulas will lead to a strong noise amplification in the projection estimate and the subsequent refractive image obtained using conventional algorithms like filtered backprojection (FBP). Tomographic reconstruction by use of statistical methods can account for the noise model and a priori information, and thereby can produce images with better quality over conventional filtered backprojection algorithms. In this work, we demonstrate an iterative image reconstruction method that exploits the second-order statistical properties of the projection data can mitigate noise amplification in PCT. The autocovariance function of the reconstructed refractive images was empirically computed and shows smaller and shorter noise correlation compared to those obtained using the FBP and unweighted penalized least-squares methods. Concepts from statistical decision theory are applied to demonstrate that the statistical properties of images produced by our method can improve signal detectability.

A fast forward projection using multithreads for multirays on GPUs in medical image reconstruction.

PURPOSE: Iterative reconstruction techniques hold great potential to mitigate the effects of data noise and/or incompleteness, and hence can facilitate the patient dose reduction. However, they are not suitable for routine clinical practice due to their long reconstruction times. In this work, the authors accelerated the computations by fully taking advantage of the highly parallel computational power on single and multiple graphics processing units (GPUs). In particular, the forward projection algorithm, which is not included in the close-form formulas, will be accelerated and optimized by using GPU here. METHODS: The main contribution is a novel forward projection algorithm that uses multithreads to handle the computations associated with a bunch of adjacent rays simultaneously. The proposed algorithm is free of divergence and bank conflict on GPU, and benefits from data locality and data reuse. It achieves the efficiency particularly by (i) employing a tiled algorithm with three-level parallelization, (ii) optimizing thread block size, (iii) maximizing data reuse on constant memory and shared memory, and (iv) exploiting built-in texture memory interpolation capability to increase efficiency. In addition, to accelerate the iterative algorithms and the Feldkamp-Davis-Kress (FDK) algorithm on GPU, the authors apply batched fast Fourier transform (FFT) to expedite filtering process in FDK and utilize projection bundling parallelism during backprojection to shorten the execution times in FDK and the expectation-maximization (EM). RESULTS: Numerical experiments conducted on an NVIDIA Tesla C1060 GPU demonstrated the superiority of the proposed algorithms in computational time saving. The forward projection, filtering, and backprojection times for generating a volume image of 512 x 512 x 512 with 360 projection data of 512 x 512 using one GPU are about 4.13, 0.65, and 2.47 s (including distance weighting), respectively. In particular, the proposed forward projection algorithm is ray-driven and its paralleli-zation strategy evolves from single-thread-for-single-ray (38.56 s), multithreads-for-single-ray (26.05 s), to multithreads-for-multirays (4.13 s). For the voxel-driven backprojection, the use of texture memory reduces the reconstruction time from 4.95 to 3.35 s. By applying the projection bundle technique, the computation time is further reduced to 2.47 s. When employing multiple GPUs, near-perfect speedups were observed as the number of GPUs increases. For example, by using four GPUs, the time for the forward projection, filtering, and backprojection are further reduced to 1.11, 0.18, and 0.66 s. The results obtained by GPU-based algorithms are virtually indistinguishable with those by CPU. CONCLUSIONS: The authors have proposed a highly optimized GPU-based forward projection algorithm, as well as the GPU-based FDK and expectation-maximization reconstruction algorithms. Our compute unified device architecture (CUDA) codes provide the exceedingly fast forward projection and backprojection that outperform those using the shading languages, cell broadband engine architecture and previous CUDA implementations. The reconstruction times in the FDK and the EM algorithms were considerably shortened, and thus can facilitate their routine usage in a variety of applications such as image quality improvement and dose reduction.

Coronary CT angiography-based estimation of myocardial perfusion territories for coronary artery FFR and wall shear stress simulation.

This study aims to apply a CCTA-derived territory-based patient-specific estimation of boundary conditions for coronary artery fractional flow reserve (FFR) and wall shear stress (WSS) simulation. The non-invasive simulation can help diagnose the significance of coronary stenosis and the likelihood of myocardial ischemia. FFR is often regarded as the gold standard to evaluate the functional significance of stenosis in coronary arteries. In another aspect, proximal wall shear stress ([Formula: see text]) can also be an indicator of plaque vulnerability. During the simulation process, the mass flow rate of the blood in coronary arteries is one of the most important boundary conditions. This study utilized the myocardium territory to estimate and allocate the mass flow rate. 20 patients are included in this study. From the knowledge of anatomical information of coronary arteries and the myocardium, the territory-based FFR and the [Formula: see text] can both be derived from fluid dynamics simulations. Applying the threshold of distinguishing between significant and non-significant stenosis, the territory-based method can reach the accuracy, sensitivity, and specificity of 0.88, 0.90, and 0.80, respectively. For significantly stenotic cases ([Formula: see text] [Formula: see text] 0.80), the vessels usually have higher wall shear stress in the proximal region of the lesion.

Noise texture and signal detectability in propagation-based x-ray phase-contrast tomography.

PURPOSE: X-ray phase-contrast tomography (PCT) is a rapidly emerging imaging modality for reconstructing estimates of an object's three-dimensional x-ray refractive index distribution. Unlike conventional x-ray computed tomography methods, the statistical properties of the reconstructed images in PCT remain unexplored. The purpose of this work is to quantitatively investigate noise propagation in PCT image reconstruction. METHODS: The authors derived explicit expressions for the autocovariance of the reconstructed absorption and refractive index images to characterize noise texture and understand how the noise properties are influenced by the imaging geometry. Concepts from statistical detection theory were employed to understand how the imaging geometry-dependent statistical properties affect the signal detection performance in a signal-known-exactly/background-known-exactly task. RESULTS: The analytical formulas for the phase and absorption autocovariance functions were implemented numerically and compared to the corresponding empirical values, and excellent agreement was found. They observed that the reconstructed refractive images are highly spatially correlated, while the absorption images are not. The numerical results confirm that the strength of the covariance is scaled by the detector spacing. Signal detection studies were conducted, employing a numerical observer. The detection performance was found to monotonically increase as the detector-plane spacing was increased. CONCLUSIONS: The authors have conducted the first quantitative investigation of noise propagation in PCT image reconstruction. The reconstructed refractive images were found to be highly spatially correlated, while absorption images were not. This is due to the presence of a Fourier space singularity in the reconstruction formula for the refraction images. The statistical analysis may facilitate the use of task-based image quality measures to further develop and optimize this emerging modality for specific applications.

Analysis of ideal observer signal detectability in phase-contrast imaging employing linear shift-invariant optical systems.

Phase-contrast imaging methods exploit variations in an object's refractive index distribution to permit the visualization of subtle features that may have very similar optical absorption properties. Although phase-contrast is often viewed as being desirable in many biomedical applications, its relative influence on signal detectability when both absorption- and phase-contrast are present remains relatively unexplored. In this work, we investigate the ideal Bayesian observer signal-to-noise ratio in phase-contrast imaging for a signal-known-exactly/background-known exactly detection task involving a weak signal. We demonstrate that this signal detectability measure can be decomposed into three contributions that have distinct interpretations associated with the imaging physics.

Effects of early ketamine exposure on cerebral gray matter volume and functional connectivity.

Ketamine has been used for medical purposes, most typically as an anesthetic, and recent studies support its use in the treatment of depression. However, ketamine tends to be abused by adolescents and young adults. In the current study, we examined the effects of early ketamine exposure on brain structure and function. We employed MRI to assess the effects of ketamine abuse on cerebral gray matter volume (GMV) and functional connectivity (FC) in 34 users and 19 non-users, employing covariates. Ketamine users were categorized as adolescent-onset and adult-onset based on when they were first exposed to ketamine. Imaging data were processed by published routines in SPM and AFNI. The results revealed lower GMV in the left precuneus in ketamine users, with a larger decrease in the adolescent-onset group. The results from a seed-based correlation analysis show that both ketamine groups had higher functional connectivity between left precuneus (seed) and right precuneus than the control group. Compared to controls, ketamine users showed decreased GMV in the right insula, left inferior parietal lobule, left dorsolateral prefrontal cortex/superior frontal gyrus, and left medial orbitofrontal cortex. These preliminary results characterize the effects of ketamine misuse on brain structure and function and highlight the influence of earlier exposure to ketamine on the development of the brain. The precuneus, a structure of central importance to cerebral functional organization, may be particularly vulnerable to the influences of early ketamine exposure. How these structural and functional brain changes may relate to the cognitive and affective deficits remains to be determined with a large cohort of participants.

Influence of imaging geometry on noise texture in quantitative in-line X-ray phase-contrast imaging.

Quantitative in-line X-ray phase-contrast imaging methods seek to reconstruct separate images that depict an object's projected absorption and refractive properties. An understanding of the statistical properties of the reconstructed images can facilitate the identification of optimal imaging parameters for specific diagnostic tasks. However, the statistical properties of quantitative X-ray phase-contrast imaging remain largely unexplored. In this work, we derive analytic expressions that describe the second-order statistics of the reconstructed absorption and phase images. Concepts from statistical decision theory are applied to demonstrate how the statistical properties of images corresponding to distinct imaging geometries can influence signal detectability.

Image reconstruction in quantitative X-ray phase-contrast imaging employing multiple measurements.

X-ray phase-contrast imaging is a technique that aims to reconstruct the projected absorption and refractive index distributions of an object. One common feature of reconstruction formulas for phase-contrast imaging is the presence of isolated Fourier domain singularities, which can greatly amplify the noise levels in the estimated Fourier domain and lead to noisy and/or distorted images in spatial domain. In this article, we develop a statistically optimal reconstruction method that employs multiple (>2) measurement states to mitigate the noise amplification effects due to singularities in the reconstruction formula. Computer-simulation studies are carried out to quantitatively and systematically investigate the developed method, within the context of propagation-based X-ray phase-contrast imaging. The reconstructed images are shown to possess dramatically reduced noise levels and greatly enhanced imaging contrast.

An extended diffraction-enhanced imaging method for implementing multiple-image radiography.

Diffraction-enhanced imaging (DEI) is an analyser-based x-ray imaging method that produces separate images depicting the projected x-ray absorption and refractive properties of an object. Because the imaging model of DEI does not account for ultra-small-angle x-ray scattering (USAXS), the images produced in DEI can contain artefacts and inaccuracies in medical imaging applications. In this work, we investigate an extended DEI method for concurrent reconstruction of three images that depict an object's projected x-ray absorption, refraction and USAXS properties. The extended DEI method can be viewed as an implementation of the recently proposed multiple-image radiography paradigm. Validation studies are conducted by use of computer-simulated and synchrotron measurement data.

Numerical modeling of nanodrug distribution in tumors with heterogeneous vasculature.

The distribution and accumulation of nanoparticle dosage in a tumor are important in evaluating the effectiveness of cancer treatment. The cell survival rate can quantify the therapeutic effect, and the survival rates after multiple treatments are helpful to evaluate the efficacy of a chemotherapy plan. We developed a mathematical tumor model based on the governing equations describing the fluid flow and particle transport to investigate the drug transportation in a tumor and computed the resulting cumulative concentrations. The cell survival rate was calculated based on the cumulative concentration. The model was applied to a subcutaneous tumor with heterogeneous vascular distributions. Various sized dextrans and doxorubicin were respectively chosen as the nanodrug carrier and the traditional chemotherapeutic agent for comparison. The results showed that: 1) the largest nanoparticle drug in the current simulations yielded the highest cumulative concentration in the well vascular region, but second lowest in the surrounding normal tissues, which implies it has the best therapeutic effect to tumor and at the same time little harmful to normal tissue; 2) on the contrary, molecular chemotherapeutic agent produced the second lowest cumulative concentration in the well vascular tumor region, but highest in the surrounding normal tissue; 3) all drugs have very small cumulative concentrations in the tumor necrotic region, where drug transport is solely through diffusion. This might mean that it is hard to kill tumor stem cells hiding in it. The current model indicated that the effectiveness of the anti-tumor drug delivery was determined by the interplay of the vascular density and nanoparticle size, which governs the drug transport properties. The use of nanoparticles as anti-tumor drug carriers is generally a better choice than molecular chemotherapeutic agent because of its high treatment efficiency on tumor cells and less damage to normal tissues.

Investigation of the spatiotemporal responses of nanoparticles in tumor tissues with a small-scale mathematical model.

The transport and accumulation of anticancer nanodrugs in tumor tissues are affected by many factors including particle properties, vascular density and leakiness, and interstitial diffusivity. It is important to understand the effects of these factors on the detailed drug distribution in the entire tumor for an effective treatment. In this study, we developed a small-scale mathematical model to systematically study the spatiotemporal responses and accumulative exposures of macromolecular carriers in localized tumor tissues. We chose various dextrans as model carriers and studied the effects of vascular density, permeability, diffusivity, and half-life of dextrans on their spatiotemporal concentration responses and accumulative exposure distribution to tumor cells. The relevant biological parameters were obtained from experimental results previously reported by the Dreher group. The area under concentration-time response curve (AUC) quantified the extent of tissue exposure to a drug and therefore was considered more reliable in assessing the extent of the overall drug exposure than individual concentrations. The results showed that 1) a small macromolecule can penetrate deep into the tumor interstitium and produce a uniform but low spatial distribution of AUC; 2) large macromolecules produce high AUC in the perivascular region, but low AUC in the distal region away from vessels; 3) medium-sized macromolecules produce a relatively uniform and high AUC in the tumor interstitium between two vessels; 4) enhancement of permeability can elevate the level of AUC, but have little effect on its uniformity while enhancement of diffusivity is able to raise the level of AUC and improve its uniformity; 5) a longer half-life can produce a deeper penetration and a higher level of AUC distribution. The numerical results indicate that a long half-life carrier in plasma and a high interstitial diffusivity are the key factors to produce a high and relatively uniform spatial AUC distribution in the interstitium.

Digital holographic microtomography for high-resolution refractive index mapping of live cells.

Quantification of three-dimensional (3D) refractive index (RI) with sub-cellular resolution is achieved by digital holographic microtomography (DHµT) using quantitative phase images measured at multiple illumination angles. The DHµT system achieves sensitive and fast phase measurements based on iterative phase extraction algorithm and asynchronous phase shifting interferometry without any phase monitoring or active control mechanism. A reconstruction algorithm, optical diffraction tomography with projection on convex sets and total variation minimization, is implemented to substantially reduce the number of angular scattered fields needed for reconstruction without sacrificing the accuracy and quality of the reconstructed 3D RI distribution. Tomogram of a living CA9-22 cell is presented to demonstrate the performance of the method. Further, a statistical analysis of the average RI of the nucleoli, the nucleus excluding the nucleoli and the cytoplasm of twenty CA9-22 cells is performed.

Accelerating image reconstruction in three-dimensional optoacoustic tomography on graphics processing units.

PURPOSE: Optoacoustic tomography (OAT) is inherently a three-dimensional (3D) inverse problem. However, most studies of OAT image reconstruction still employ two-dimensional imaging models. One important reason is because 3D image reconstruction is computationally burdensome. The aim of this work is to accelerate existing image reconstruction algorithms for 3D OAT by use of parallel programming techniques. METHODS: Parallelization strategies are proposed to accelerate a filtered backprojection (FBP) algorithm and two different pairs of projection/backprojection operations that correspond to two different numerical imaging models. The algorithms are designed to fully exploit the parallel computing power of graphics processing units (GPUs). In order to evaluate the parallelization strategies for the projection/backprojection pairs, an iterative image reconstruction algorithm is implemented. Computer simulation and experimental studies are conducted to investigate the computational efficiency and numerical accuracy of the developed algorithms. RESULTS: The GPU implementations improve the computational efficiency by factors of 1000, 125, and 250 for the FBP algorithm and the two pairs of projection/backprojection operators, respectively. Accurate images are reconstructed by use of the FBP and iterative image reconstruction algorithms from both computer-simulated and experimental data. CONCLUSIONS: Parallelization strategies for 3D OAT image reconstruction are proposed for the first time. These GPU-based implementations significantly reduce the computational time for 3D image reconstruction, complementing our earlier work on 3D OAT iterative image reconstruction.

Accelerating image reconstruction in dual-head PET system by GPU and symmetry properties.

Positron emission tomography (PET) is an important imaging modality in both clinical usage and research studies. We have developed a compact high-sensitivity PET system that consisted of two large-area panel PET detector heads, which produce more than 224 million lines of response and thus request dramatic computational demands. In this work, we employed a state-of-the-art graphics processing unit (GPU), NVIDIA Tesla C2070, to yield an efficient reconstruction process. Our approaches ingeniously integrate the distinguished features of the symmetry properties of the imaging system and GPU architectures, including block/warp/thread assignments and effective memory usage, to accelerate the computations for ordered subset expectation maximization (OSEM) image reconstruction. The OSEM reconstruction algorithms were implemented employing both CPU-based and GPU-based codes, and their computational performance was quantitatively analyzed and compared. The results showed that the GPU-accelerated scheme can drastically reduce the reconstruction time and thus can largely expand the applicability of the dual-head PET system.

Analyzer-based phase-contrast x-ray imaging of carotid plaque microstructure.

BACKGROUND: Plaque vulnerability depends, in part, on composition. Imaging techniques are needed that can aid the prediction of plaque stability. High-contrast images of soft-tissue structure have been obtained with x-ray phase-contrast (PC) imaging. This research investigates multiple image radiography (MIR), an x-ray PC imaging technique, for evaluation of human carotid artery plaques. METHODS: Carotid plaques were imaged with ultrasound and subsequently excised and formalin fixed. MIR imaging was performed. By using synchrotron radiation, conventional radiographs were acquired for comparison. Image texture measures were computed for soft-tissue regions of the plaques. RESULTS: Ultrasound evaluation identified plaques as homogeneous without calcifications. MIR images revealed complex heterogeneous structure with multiple microcalcifications consistent with histology, and possessed more image texture in specific regions than conventional radiographs (P < .05). MIR refraction images allowed imaging of the geometric structure of tissue interfaces within the plaques, while scatter images contained more texture in soft-tissue regions than absorption or refraction images. CONCLUSIONS: X-ray PC imaging better depicts plaque soft-tissue heterogeneity than ultrasound or conventional radiographs. MIR imaging technique should be investigated further as a viable imaging technique to identify high-risk plaques.

Ultrasound sonication with microbubbles disrupts blood vessels and enhances tumor treatments of anticancer nanodrug.

Ultrasound (US) sonication with microbubbles (MBs) has the potential to disrupt blood vessels and enhance the delivery of drugs into the sonicated tissues. In this study, mouse ear tumors were employed to investigate the therapeutic effects of US, MBs, and pegylated liposomal doxorubicin (PLD) on tumors. Tumors started to receive treatments when they grew up to about 15 mm(3) (early stage) with injection of PLD 10 mg/kg, or up to 50 mm(3) (medium stage) with PLD 6 (or 4) mg/kg. Experiments included the control, PLD alone, PLD + MBs + US, US alone, and MBs + US groups. The procedure for the PLD + MBs + US group was that PLD was injected first, MB (SonoVue) injection followed, and then US was immediately sonicated on the tumor. The results showed that: (1) US sonication with MBs was always able to produce a further hindrance to tumor growth for both early and medium-stage tumors; (2) for the medium-stage tumors, 6 mg/kg PLD alone was able to inhibit their growth, while it did not work for 4 mg/kg PLD alone; (3) with the application of MBs + US, 4 mg/kg PLD was able to inhibit the growth of medium-stage tumors; (4) for early stage tumors after the first treatment with a high dose of PLD alone (10 mg/kg), the tumor size still increased for several days and then decreased (a biphasic pattern); (5) MBs + US alone was able to hinder the growth of early stage tumors, but unable to hinder that of medium stage tumors. The results of histological examinations and blood perfusion measurements indicated that the application of MBs + US disrupts the tumor blood vessels and enhances the delivery of PLD into tumors to significantly inhibit tumor growth.