Biomedical subjects

Xuejun Gu

Publications and source records attributed to Xuejun Gu.

8 recordsLinked to original sources

PubMed2026

Artificial intelligence (AI) uses in stereotactic radiosurgery (SRS): diagnosis with brain metastasis (BM) - A systematic review.

BACKGROUND: Brain metastases (BM) are the most common intracranial tumors in adults, and stereotactic radiosurgery (SRS) has become a mainstay of management. However, several diagnostic challenges persist in the SRS pathway, particularly the differentiation of radiation necrosis (RN) from true tumor progression, which conventional MRI and even advanced imaging techniques often cannot reliably resolve. Recent advances in artificial intelligence (AI) offer the potential to address these diagnostic limitations. This systematic review synthesizes current literature on AI applications for MRI-based diagnostic decision support in BM patients undergoing SRS, with a focus on radiomics and deep learning tools for distinguishing RN from progression, classifying molecular and histologic subtypes, and predicting treatment response. METHODS: A systematic review was performed in accordance with PRISMA guidelines. PubMed, Web of Science, and Scopus were searched using a targeted query combining terms related to AI, brain metastasis, diagnosis or imaging, and SRS. After screening 483 records and applying strict inclusion and exclusion criteria, 18 studies published between 2015 and 2025 were included. Data were extracted on study design, cohort characteristics, imaging modality, AI methodology, validation strategy, and reported diagnostic performance. RESULTS: Among the 18 included studies, AI models demonstrated strong performance across diagnostic tasks in the BM-SRS pathway. The differentiation of RN from true tumor progression was the most extensively studied application, addressed by 14 of 18 studies, with reported AUCs ranging from 0.71 to 0.94. Support vector machines, random-forest ensembles, convolutional neural networks, and transformer-based multimodal architectures were widely used. The literature evolved from single-sequence radiomic classifiers in 2018 to multimodal deep learning frameworks fusing imaging with clinical and genomic data in 2025. Contrast-enhanced T1-weighted MRI was the dominant imaging input, and texture-based radiomic features (GLCM, GLSZM, GLDM, and wavelet-derived features) were the most consistently predictive. The highest-performing models reached AUCs of 0.85-0.91 through multimodal integration of imaging with clinical and genomic features, and consistently outperformed expert neuroradiologist read on matched cases. Remaining studies addressed longitudinal segmentation-based detection of local failure and adverse radiation effects, BRAF mutation status in melanoma BM, early Gamma Knife treatment response, and primary tumor histology classification, with more variable performance. CONCLUSION: AI models, particularly those integrating MRI-derived radiomic features with clinical and genomic data, show high accuracy in supporting diagnostic decisions for BM patients treated with SRS. The post-SRS differentiation of radiation necrosis from true tumor progression has reached the greatest level of maturity and is closest to clinical translation, with potential to reduce unnecessary biopsies, personalize surveillance intervals, and rationalize treatment-pathway decisions. Other diagnostic applications, including molecular subtyping and primary tumor histology classification, remain exploratory and require further multicenter validation. Integration of AI tools into multidisciplinary tumor-board workflows, combined with prospective validation and standardized reporting, will be essential to realize the full clinical benefits of AI in SRS for brain metastases.

Humans↗

PubMed2006

Spatially varying optical and acoustic property reconstruction using finite-element-based photoacoustic tomography.

A finite-element reconstruction algorithm for simultaneous reconstruction of both optical and acoustic properties of heterogeneous media is presented. The algorithm is based on the Helmholtz-like photoacoustic wave equation in the frequency domain. A dual meshing scheme is described and an adjoint sensitivity method is adopted for efficient inverse computation. The algorithm is implemented with the second-order absorbing boundary conditions and with a multireceiving and multifrequency strategy. The algorithm is evaluated using simulated data under various practical cases including different noise levels, varied range of receiving frequency, different contrast levels between the heterogeneity and background region, and multiple targets. The effect of acoustic heterogeneity on conventional pure optical absorption reconstruction is also studied.

Acoustics↗

PubMed2005

Model-based ultrasound tomography: tissue phantom experiments.

We present a detailed experimental study to evaluate our finite element based nonlinear reconstruction algorithm for recovery of acoustic properties in heterogeneous scattering media. Using a circularly scanning ultrasound system at 500 KHz, tissue phantom experiments were performed to study spatial resolution and contrast issues in model-based ultrasound tomography. Our results show that both acoustic attenuation and speed images can be quantitatively reconstructed in terms of the location, size, shape, and acoustic property value of the target when different contrast levels between the target and background were used. We also demonstrate that a high contrast target as small as 3 mm in diameter can be quantitatively resolved with our acoustic speed and attenuation images.

Algorithms↗

PubMed2004

Differentiation of cysts from solid tumors in the breast with diffuse optical tomography.

RATIONALE AND OBJECTIVES: Near-infrared diffuse optical tomography (DOT) is an emerging imaging technology that has the potential to offer enhanced contrast resolution over the existing technologies for detection and diagnosis of breast cancer. Thus far, the clinical evaluation of DOT has been largely limited to solid tumors. A pilot clinical study focused on DOT imaging of breasts with cysts is presented. MATERIALS AND METHODS: Six cases were studied using the recently developed compact, parallel-detection DOT system. Images characterizing the tissue absorption and scattering were obtained with a finite element-based reconstruction algorithm. The optical images were compared with the mammograms and sonograms. In one case, in vitro measurements of optical properties were conducted for the fluid obtained from needle aspiration. RESULTS: Substantial contrast between cyst and adjacent parenchyma is observed. For the six cases evaluated, the locations and sizes of cysts imaged optically are accurate and consistent with the mammographic and sonographic findings. For the case that aspiration was performed, the absorption and scattering coefficients imaged in the cyst region are quantitatively accurate compared with that measured in vitro from the fluid aspirated. CONCLUSION: This pilot study shows that cysts ranging from 1-4 cm in diameter can be quantitatively imaged. They can be differentiated from solid breast tumors because cysts generally demonstrate lower absorption and scattering coefficients compared with the surrounding normal tissue, whereas solid tumors show concurrent higher absorption and scattering related to the normal tissue.

Adult↗

PubMed2003

In vivo breast imaging with diffuse optical tomography based on higher-order diffusion equations.

We report on in vivo absorption and scattering imaging of a human breast cyst and implant, using a reconstruction algorithm based on our third-order diffusion equations. To validate these in vivo images, a series of phantom experiments were conducted, in which we used low-absorbing and low-scattering heterogeneities to mimic a breast cyst or implant. These heterogeneities or targets were composed of pure water or a mixture of water and very dilute Intralipid (0.05% and 0.1%). The phantom experiment confirmed the quantitative imaging capability of our improved algorithm for reconstructing heterogeneities where the conventional diffusion approximation is inadequate. Pilot clinical results from female volunteers indicate that enhanced diffuse optical tomography can quantitatively image findings such as breast cysts or implants in which the absorption and scattering coefficients are usually low.

Adult↗

PubMed2003

Mesh-based enhancement schemes in diffuse optical tomography.

Two mesh-based methods including dual meshing and adaptive meshing are developed to improve the finite element-based reconstruction of both absorption and scattering images of heterogeneous turbid media. The idea of dual meshing scheme is to use a fine mesh for the solution of photon propagation and a coarse mesh for the inversion of optical property distributions. The adaptive meshing method is accomplished by the automatic mesh refinement in the region of heterogeneity during reconstruction. These schemes are validated using tissue-like phantom measurements. Our results demonstrate the capabilities of the dual meshing and adaptive meshing in both qualitative and quantitative improvement of optical image reconstruction.

Algorithms↗

PubMed2002

Absorption and scattering images of heterogeneous scattering media can be simultaneously reconstructed by use of dc data.

We present a carefully designed phantom experimental study aimed to provide solid evidence that both absorption and scattering images of heterogeneous scattering media can be reconstructed independently from dc data. We also study the important absorption-scattering cross-talk issue. In this regard, we develop a simple normalizing scheme that is incorporated into our nonlinear finite-element-based reconstruction algorithm. Our results from the controlled phantom experiments show that the cross talk of an absorption object appearing in scattering images can be eliminated and that the cross talk of a scattering object appearing in absorption images can be reduced considerably. In addition, these carefully designed phantom experiments clearly suggest that both absorption and scattering images can be simultaneously recovered and quantitatively separated in highly scattering media by use of dc measurements. Finally, we discuss our results in light of recent theoretical findings on nonuniqueness for dc image reconstruction.

Absorption↗

PubMedDate not supplied

Strategies for improved dCO2 removal in large-scale fed-batch cultures.

Carbon dioxide buildup in large-scale reactors can be detrimental to cell growth and productivity. In case of protein X, a therapeutic glycoprotein, when cultures were scaled up from bench scale to the pilot plant, there was a 40% loss of specific productivity. The dissolved CO(2) (dCO(2)) level was 179 +/- 9 mmHg at the pilot plant scale and 68 +/- 13 mmHg at bench scale. The authors proposed a comprehensive approach to maintain dCO(2) levels between 40 and 120 mmHg throughout the 14-day fed-batch process. A cell-free experiment was used to investigate the impact of the following parameters on dCO(2) removal: (1) sparge rate, (2) agitator speed, (3) bubble size, (4) bicarbonate concentration, (5) impeller position, and (6) aeration rate at the headspace of bioreactor. dCO(2) was measured using a fiber optic based probe. dCO(2) removal rate was a strong function of sparge rate and a weak function of agitator speed. Bubble size was modulated by the presence or absence of a sparge stone (10 microm pore size, 1 cm pipe i.d.). Open pipe provided 3- to 4-fold better dCO(2) removal for the same mass transfer coefficient (k(L)a) value. A mathematical model and a bench-scale experiment indicated that the benefit of a lower level of sodium bicarbonate in the culture medium was transient for batch and fed-batch cultures. Thus, this strategy was not used at pilot scale. Decreasing top impeller position improved k(L)a of dCO(2) by 2-fold. Changing headspace aeration rate from 0.02 to 0.04 vvm had no impact on dCO(2) removal. Two pilot runs were conducted using (A) open pipe and (B) antifoam in the presence of sparge stone, both in conjunction with lower impeller position. The presence of antifoam may interfere in product purification; however, demonstration of antifoam removal can be difficult. Open pipe allowed an alternative to using antifoam, as foam level with open pipe was significantly less. Both strategies successfully reduced dCO(2) level by 2.5-fold (179 +/- 9 vs 72 +/- 9 mmHg). Titer at day 10 of culture improved by 1.5-fold. Specific productivity improved by 41%. Historically, cultures were harvested around day 9-11 because of the high amount of foam; both strategies allowed the cultures to be extended up to day 14, resulting in 2-fold higher titer compared to that of the historical control without compromising protein quality.

Bioreactors↗