Adaptive and Iterative Learning with Multi-perspective Regularizations for Metal Artifact Reduction.

Metal artifact reduction (MAR) is important for clinical diagnosis with CT images. The existing state-of-the-art deep learning methods usually suppress metal artifacts in sinogram or image domains or both. However, their performance is limited by the inherent characteristics of the two domains, i.e., the errors introduced by local manipulations in the sinogram domain would propagate throughout the whole image during backprojection and lead to serious secondary artifacts, while it is difficult to distinguish artifacts from actual image features in the image domain. To alleviate these limitations, this study analyzes the desirable properties of wavelet transform in-depth and proposes to perform MAR in the wavelet domain. First, wavelet transform yields components that possess spatial correspondence with the image, thereby preventing the spread of local errors to avoid secondary artifacts. Second, using wavelet transform could facilitate identification of artifacts from image since metal artifacts are mainly high-frequency signals. Taking these advantages of the wavelet transform, this paper decomposes an image into multiple wavelet components and introduces multi-perspective regularizations into the proposed MAR model. To improve the transparency and validity of the model, all the modules in the proposed MAR model are designed to reflect their mathematical meanings. In addition, an adaptive wavelet module is also utilized to enhance the flexibility of the model. To optimize the model, an iterative algorithm is developed. The evaluation on both synthetic and real clinical datasets consistently confirms the superior performance of the proposed method over the competing methods.

Constructing hierarchical attentive functional brain networks for early AD diagnosis.

Analyzing functional brain networks (FBN) with deep learning has demonstrated great potential for brain disorder diagnosis. The conventional construction of FBN is typically conducted at a single scale with a predefined brain region atlas. However, numerous studies have identified that the structure and function of the brain are hierarchically organized in nature. This urges the need of representing FBN in a hierarchical manner for more effective analysis of the complementary diagnostic insights at different scales. To this end, this paper proposes to build hierarchical FBNs adaptively within the Transformer framework. Specifically, a sparse attention-based node-merging module is designed to work alongside the conventional network feature extraction modules in each layer. The proposed module generates coarser nodes for further FBN construction and analysis by combining fine-grained nodes. By stacking multiple such layers, a hierarchical representation of FBN can be adaptively learned in an end-to-end manner. The hierarchical structure can not only integrate the complementary information from multiscale FBN for joint analysis, but also reduce the model complexity due to decreasing node sizes. Moreover, this paper argues that the nodes defined by the existing atlases are not necessarily the optimal starting level to build FBN hierarchy and exploring finer nodes may further enrich the FBN representation. In this regard, each predefined node in an atlas is split into multiple sub-nodes, overcoming the scale limitation of the existing atlases. Extensive experiments conducted on various data sets consistently demonstrate the superior performance of the proposed method over the competing methods.

Multi-channel Optimization Generative Model for Stable Ultra-Sparse-View CT Reconstruction.

Score-based generative model (SGM) has risen to prominence in sparse-view CT reconstruction due to its impressive generation capability. The consistency of data is crucial in guiding the reconstruction process in SGM-based reconstruction methods. However, the existing data consistency policy exhibits certain limitations. Firstly, it employs partial data from the reconstructed image of iteration process for image updates, which leads to secondary artifacts with compromising image quality. Moreover, the updates to the SGM and data consistency are considered as distinct stages, disregarding their interdependent relationship. Additionally, the reference image used to compute gradients in the reconstruction process is derived from intermediate result rather than ground truth. Motivated by the fact that a typical SGM yields distinct outcomes with different random noise inputs, we propose a Multi-channel Optimization Generative Model (MOGM) for stable ultra-sparse-view CT reconstruction by integrating a novel data consistency term into the stochastic differential equation model. Notably, the unique aspect of this data consistency component is its exclusive reliance on original data for effectively confining generation outcomes. Furthermore, we pioneer an inference strategy that traces back from the current iteration result to ground truth, enhancing reconstruction stability through foundational theoretical support. We also establish a multi-channel optimization reconstruction framework, where conventional iterative techniques are employed to seek the reconstruction solution. Quantitative and qualitative assessments on 23 views datasets from numerical simulation, clinical cardiac and sheep's lung underscore the superiority of MOGM over alternative methods. Reconstructing from just 10 and 7 views, our method consistently demonstrates exceptional performance.

Spectrum learning for super-resolution tomographic reconstruction.

Objective. Computed Tomography (CT) has been widely used in industrial high-resolution non-destructive testing. However, it is difficult to obtain high-resolution images for large-scale objects due to their physical limitations. The objective is to develop an improved super-resolution technique that preserves small structures and details while efficiently capturing high-frequency information.Approach. The study proposes a new deep learning based method called spectrum learning (SPEAR) network for CT images super-resolution. This approach leverages both global information in the image domain and high-frequency information in the frequency domain. The SPEAR network is designed to reconstruct high-resolution images from low-resolution inputs by considering not only the main body of the images but also the small structures and other details. The symmetric property of the spectrum is exploited to reduce weight parameters in the frequency domain. Additionally, a spectrum loss is introduced to enforce the preservation of both high-frequency components and global information.Main results. The network is trained using pairs of low-resolution and high-resolution CT images, and it is fine-tuned using additional low-dose and normal-dose CT image pairs. The experimental results demonstrate that the proposed SPEAR network outperforms state-of-the-art networks in terms of image reconstruction quality. The approach successfully preserves high-frequency information and small structures, leading to better results compared to existing methods. The network's ability to generate high-resolution images from low-resolution inputs, even in cases of low-dose CT images, showcases its effectiveness in maintaining image quality.Significance. The proposed SPEAR network's ability to simultaneously capture global information and high-frequency details addresses the limitations of existing methods, resulting in more accurate and informative image reconstructions. This advancement can have substantial implications for various industries and medical diagnoses relying on accurate imaging.

Wavelet-Inspired Multi-channel Score-based Model for Limited-angle CT Reconstruction.

Score-based generative model (SGM) has demonstrated great potential in the challenging limited-angle CT (LA-CT) reconstruction. SGM essentially models the probability density of the ground truth data and generates reconstruction results by sampling from it. Nevertheless, direct application of the existing SGM methods to LA-CT suffers multiple limitations. Firstly, the directional distribution of the artifacts attributing to the missing angles is ignored. Secondly, the different distribution properties of the artifacts in different frequency components have not been fully explored. These drawbacks would inevitably degrade the estimation of the probability density and the reconstruction results. After an in-depth analysis of these factors, this paper proposes a Wavelet-Inspired Score-based Model (WISM) for LA-CT reconstruction. Specifically, besides training a typical SGM with the original images, the proposed method additionally performs the wavelet transform and models the probability density in each wavelet component with an extra SGM. The wavelet components preserve the spatial correspondence with the original image while performing frequency decomposition, thereby keeping the directional property of the artifacts for further analysis. On the other hand, different wavelet components possess more specific contents of the original image in different frequency ranges, simplifying the probability density modeling by decomposing the overall density into component-wise ones. The resulting two SGMs in the image-domain and wavelet-domain are integrated into a unified sampling process under the guidance of the observation data, jointly generating high-quality and consistent LA-CT reconstructions. The experimental evaluation on various datasets consistently verifies the superior performance of the proposed method over the competing method.

A rapid localization and analysis method for isoquinoline alkaloids with fluorescence in Coptis chinensis Franch. By fabricating the nano-silver sol as a substrate for surface-enhanced Raman spectroscopy.

BACKGROUND: The quality of traditional Chinese medicines (TCMs) directly impacts their clinical efficacy and drug safety, making standardization a critical component of modern TCMs. Surface-enhanced Raman spectroscopy (SERS) is an effective physical detection method with speed, sensitivity, and suitability for large sample analyses. In this study, a SERS analysis method was developed using a nano-silver sol as the matrix to address the interference of fluorescence components in TCMs and overcome the limitations of traditional detection methods. RESULTS: The higher sensitivity and efficiency of SERS was used, enabling detection of a single sample within 30 s. Coptis chinensis Franch. (CCF) was chosen as the model medicine, the nano-silver sol was used as the matrix, and CCF's fourteen main fluorescent alkaloids were tested as index components. Typical signal peaks of the main components in CCF corresponded to the bending deformation of the nitrogen-containing ring plane outer ring system, methoxy stretching vibration, and isoquinoline ring deformation vibration. Through SERS detection of different parts, the distribution content of the main active components in the cortex of CCF was found to be lower than that in the xylem and phloem. Additionally, rapid quality control analyses indicated that among the nine batches of original medicinal materials purchased from Emei and Guangxi, the main active ingredient showed a higher content. SIGNIFICANCE: A SERS-based method for the rapid localization and analysis of multiple components of TCMs was established. The findings highlight the potential of SERS as a valuable tool for the analysis and quality control of TCMs, especially for fluorescent components.

Two-and-a-half order score-based model for solving 3D ill-posed inverse problems.

Computed Tomography (CT) and Magnetic Resonance Imaging (MRI) are crucial technologies in the field of medical imaging. Score-based models demonstrated effectiveness in addressing different inverse problems encountered in the field of CT and MRI, such as sparse-view CT and fast MRI reconstruction. However, these models face challenges in achieving accurate three dimensional (3D) volumetric reconstruction. The existing score-based models predominantly concentrate on reconstructing two-dimensional (2D) data distributions, resulting in inconsistencies between adjacent slices in the reconstructed 3D volumetric images. To overcome this limitation, we propose a novel two-and-a-half order score-based model (TOSM). During the training phase, our TOSM learns data distributions in 2D space, simplifying the training process compared to working directly on 3D volumes. However, during the reconstruction phase, the TOSM utilizes complementary scores along three directions (sagittal, coronal, and transaxial) to achieve a more precise reconstruction. The development of TOSM is built on robust theoretical principles, ensuring its reliability and efficacy. Through extensive experimentation on large-scale sparse-view CT and fast MRI datasets, our method achieved state-of-the-art (SOTA) results in solving 3D ill-posed inverse problems, averaging a 1.56 dB peak signal-to-noise ratio (PSNR) improvement over existing sparse-view CT reconstruction methods across 29 views and 0.87 dB PSNR improvement over existing fast MRI reconstruction methods with × 2 acceleration. In summary, TOSM significantly addresses the issue of inconsistency in 3D ill-posed problems by modeling the distribution of 3D data rather than 2D distribution which has achieved remarkable results in both CT and MRI reconstruction tasks.

Wavelet-Improved Score-Based Generative Model for Medical Imaging.

The score-based generative model (SGM) has demonstrated remarkable performance in addressing challenging under-determined inverse problems in medical imaging. However, acquiring high-quality training datasets for these models remains a formidable task, especially in medical image reconstructions. Prevalent noise perturbations or artifacts in low-dose Computed Tomography (CT) or under-sampled Magnetic Resonance Imaging (MRI) hinder the accurate estimation of data distribution gradients, thereby compromising the overall performance of SGMs when trained with these data. To alleviate this issue, we propose a wavelet-improved denoising technique to cooperate with the SGMs, ensuring effective and stable training. Specifically, the proposed method integrates a wavelet sub-network and the standard SGM sub-network into a unified framework, effectively alleviating inaccurate distribution of the data distribution gradient and enhancing the overall stability. The mutual feedback mechanism between the wavelet sub-network and the SGM sub-network empowers the neural network to learn accurate scores even when handling noisy samples. This combination results in a framework that exhibits superior stability during the learning process, leading to the generation of more precise and reliable reconstructed images. During the reconstruction process, we further enhance the robustness and quality of the reconstructed images by incorporating regularization constraint. Our experiments, which encompass various scenarios of low-dose and sparse-view CT, as well as MRI with varying under-sampling rates and masks, demonstrate the effectiveness of the proposed method by significantly enhanced the quality of the reconstructed images. Especially, our method with noisy training samples achieves comparable results to those obtained using clean data. Our code at https://zenodo.org/record/8266123.

Dataset-Driven Unsupervised Object Discovery for Region-Based Instance Image Retrieval.

Instance image retrieval could greatly benefit from discovering objects in the image dataset. This not only helps produce more reliable feature representation but also better informs users by delineating query-matched object regions. However, object classes are usually not predefined in a retrieval dataset and class label information is generally unavailable in image retrieval. This situation makes object discovery a challenging task. To address this, we propose a novel dataset-driven unsupervised object discovery framework. By utilizing deep feature representation and weakly-supervised object detection, we explore supervisory information from within an image dataset, construct class-wise object detectors, and assign multiple detectors to each image for detection. To efficiently construct object detectors for large image datasets, we propose a novel "base-detector repository" and derive a fast way to generate the base detectors. In addition, the whole framework is designed to work in a self-boosting manner to iteratively refine object discovery. Compared with existing unsupervised object detection methods, our framework produces more accurate object discovery results. Different from supervised detection, we need neither manual annotation nor auxiliary datasets to train object detectors. Experimental study demonstrates the effectiveness of the proposed framework and the improved performance for region-based instance image retrieval.

Spectral CT reconstruction via Spectral-Image Tensor and Bidirectional Image-gradient minimization.

It is challenging to obtain good image quality in spectral computed tomography (CT) as the photon-number for the photon-counting detectors is limited for each narrow energy bin. This results in a lower signal to noise ratio (SNR) for the projections. To handle this issue, we first formulate the weight bidirectional image gradient with L0-norm constraint of spectral CT image. Then, as a new regularizer, bidirectional image gradient with L0-norm constraint is introduced into the tensor decomposition model, generating the Spectral-Image Tensor and Bidirectional Image-gradient Minimization (SITBIM) algorithm. Finally, the split-Bregman method is employed to optimize the proposed SITBIM mathematical model. The experiments on the numerical mouse phantom and real mouse experiments are designed to validate and evaluate the SITBIM method. The results demonstrate that the SITBIM can outperform other state-of-the-art methods (including TVM, TV + LR, SSCMF and NLCTF). INDEX TERMS: -spectral CT, image reconstruction, tensor decomposition, unidirectional image gradient, image similarity.

Diffusion Kernel Attention Network for Brain Disorder Classification.

Constructing and analyzing functional brain networks (FBN) has become a promising approach to brain disorder classification. However, the conventional successive construct-and-analyze process would limit the performance due to the lack of interactions and adaptivity among the subtasks in the process. Recently, Transformer has demonstrated remarkable performance in various tasks, attributing to its effective attention mechanism in modeling complex feature relationships. In this paper, for the first time, we develop Transformer for integrated FBN modeling, analysis and brain disorder classification with rs-fMRI data by proposing a Diffusion Kernel Attention Network to address the specific challenges. Specifically, directly applying Transformer does not necessarily admit optimal performance in this task due to its extensive parameters in the attention module against the limited training samples usually available. Looking into this issue, we propose to use kernel attention to replace the original dot-product attention module in Transformer. This significantly reduces the number of parameters to train and thus alleviates the issue of small sample while introducing a non-linear attention mechanism to model complex functional connections. Another limit of Transformer for FBN applications is that it only considers pair-wise interactions between directly connected brain regions but ignores the important indirect connections. Therefore, we further explore diffusion process over the kernel attention to incorporate wider interactions among indirectly connected brain regions. Extensive experimental study is conducted on ADHD-200 data set for ADHD classification and on ADNI data set for Alzheimer's disease classification, and the results demonstrate the superior performance of the proposed method over the competing methods.

BmNPV-induced hormone metabolic disorder in silkworm leads to enhanced locomotory behavior.

Many parasites alter the host locomotory behaviors in a way that increases their fitness and progeny transmission. Baculoviruses can manipulate host physiology and alter the locomotory behavior by inducing 'hyperactivity' (increased locomotion) or 'tree-top disease' (climbing high up to the top before dying). However, the detailed molecular mechanism underlying virus-induced this hyperactive behavior remains elusive. In the present study, we showed that BmNPV invaded into silkworm brain tissue, resulting in severe brain damage. Moreover, BmNPV infection disturbed the insect hormone balance. The content of 20-hydroxyecdysone (20E) in hemolymph was much lower during the hyperactive stage, while the dopamine (DA) titer was higher than mock infection. Exogenous hormone treatment assays demonstrated that 20E inhibits virus-induced ELA (enhanced locomotory activity), while dopamine stimulates this behavior. More specificity, injection of dopamine or its agonist promote this hyperactive behavior in BmNPV-infected larvae. Taking together, our findings revealed the important role of hormone metabolism in BmNPV-induced ELA.

Identification of A functional region in Bombyx mori nucleopolyhedrovirus VP39 that is essential for nuclear actin polymerization.

Nuclear actin polymerization plays an indispensable role in the nuclear assembly of baculovirus nucleocapsid, but the underlying viral infection-mediated mechanism remains unclear. VP39 is the major protein in baculovirus capsid, which builds the skeleton of the capsid tubular structure. VP39 is suggested in previous studies to interact with cellular actin and mediate actin polymerization. However, it is unclear about the role of VP39 in mediating nuclear actin polymerization. Results in this study indicated that vp39 deletion abolished nuclear actin polymerization, which was recovered after vp39 repair, revealing the essential part of VP39 in nuclear actin polymerization. Furthermore, a series of mutants with vp39 deletions were constructed to analyze the important region responsible for nuclear actin polymerization. In addition, intracellular localization analysis demonstrated that the amino acids 192-286 in VP39 C-terminal are responsible for nuclear actin polymerization.

Bombyx mori nucleopolyhedrovirus Bm46 is essential for efficient production of infectious BV and nucleocapsid morphogenesis.

Bombyx mori nucleopolyhedrovirus (BmNPV) orf46 (Bm46), the orthologues of Autographa californica multiple nucleopolyhedrovirus (AcMNPV) ac57, is a highly conserved gene in group Ⅰ and group Ⅱ nucleopolyhedroviruses (NPVs). However, its function in viral life cycle is unclear. Our results indicated that Bm46 transcript was detected from infected cells at 12 h post infection, while Bm46 protein was detectable from 24 to 72 h post infection. Upon the deletion of Bm46, fewer infectious BVs were produced by titer assays, but neither viral DNA synthesis nor occlusion bodies (OBs) production was affected. Electron microscopy revealed that Bm46 knockout interrupted nucleocapsid assembly and occlusion-derived virus (ODV) embedding, resulting in aberrant capsid-like tubular structures accumulated in the RZ (ring zone). Interestingly, this abnormally elongated capsid structures were consistent with the immunofluorescence microscopy results showing that VP39 assembled into long filaments and cables in the RZ. Moreover, DNA copies decreased by 30 % in occlusion bodies (OBs) produced by Bm46-knockout virus. qRT-PCR and Western blot analysis showed that the expression of VP39 was affected by Bm46 disruption. Taken together, our findings clearly pointed out that Bm46 played an important role in BV production and the proper formation of nucleocapsid morphogenesis.

Dynamic chromatin accessibility profiling reveals changes in host genome organization in response to baculovirus infection.

DNA viruses can hijack and manipulate the host chromatin state to facilitate their infection. Multiple lines of evidences reveal that DNA virus infection results in the host chromatin relocation, yet there is little known about the effects of viral infection on the architecture of host chromatin. Here, a combination of epigenomic, transcriptomic and biochemical assays were conducted to investigate the temporal dynamics of chromatin accessibility in response to Bombyx mori nucleopolyhedrovirus (BmNPV) infection. The high-quality ATAC-seq data indicated that progressive chromatin remodeling took place following BmNPV infection. Viral infection resulted in a more open chromatin architecture, along with the marginalization of host genome and nucleosome disassembly. Moreover, our results revealed that chromatin accessibility in uninfected cells was regulated by euchromatic modifications, whereas the viral-induced highly accessible chromatin regions were originally associated with facultative heterochromatic modification. Overall, our findings illustrate for the first time the organization and accessibility of host chromatin in BmNPV-infected cells, which lay the foundation for future studies on epigenomic regulation mediated by DNA viruses.

Bombyx mori nucleopolyhedrovirus utilizes a clathrin and dynamin dependent endocytosis entry pathway into BmN cells.

Bombyx mori nucleopolyhedrovirus (BmNPV) is a leading cause of silkworm mortality and economic loss to sericulture. The entry of BmNPV budded virus (BV) into host cells is a fundamental process required for the initiation of infection. However, our understanding of the mechanism of virus entry is limited and it is unclear whether BV enter BmN cells via clathrin-mediated endocytosis. In this study, we found that BV enter BmN cells through a low-pH-dependent endocytosis pathway. Inhibition assays, transmission electron microscopy (TEM) analysis, and small interfering RNAs (siRNAs) knockdown assays revealed that BV entry into BmN cells is mediated by clathrin-dependent endocytosis. Moreover, after treated with dynasore, an inhibitor of dynamin, BmNPV entry was markedly reduced, indicating that dynamin also participates in the efficient internalization of BmNPV. In addition, suppression of Rab5, Rab7 or Rab11 through siRNAs demonstrated that BV requires early and late endosomes for endocytosis in infection of BmN cells. Taken together, BmNPV uses a clathrin- and dynamin-mediated endocytic pathway into BmN cells that requires participation of Rab5 and Rab7 but not Rab11.

Networks of protein-protein interactions among structural proteins of budded virus of Bombyx mori nucleopolyhedrovirus.

The structural proteins of baculovirus are well studied, but the interactions between them remain unclear. In order to reveal protein-protein interactions among viral structural proteins and their associated proteins of the budded virus of Bombyx mori nucleopolyhedrovirus (BmNPV), the yeast two hybrid (Y2H) system was used to evaluate the interactions of 27 viral genes products. Fifty-seven interactions were identified with 51 binary interactions and 6 self-associations. Among them, 10 interactions were further confirmed by co-immunoprecipitation assays. Five interaction networks were formed based on the direct-cross Y2H assays. VP39, 38 K, and FP were identified to interact with most of the viral proteins, and may form major structural elements of the viral architecture. In addition, each envelope protein was detected to interact with more than one capsid protein. These results suggest how viral structural and structural associated proteins may assemble to form a complete virus through interacting with each other.

Identification of a novel host protein SINAL10 interacting with GP64 and its role in Bombyx mori nucleopolyhedrovirus infection.

Bombyx mori nucleopolyhedrovirus (BmNPV) is the most important pathogen of Bombyx mori, silkworm and causes severe losses in the silk industry. During the virus infectious cycle, budded virus (BVs) and occlusion-derived virus (ODVs) particles, which have identical genetic content but different phenotypes, are produced. The envelope glycoprotein GP64, specific in BVs, is involved in host cell receptor binding and is sufficient to mediate membrane fusion during the viral entry. However, the host cell factors, interacting with GP64 to mediate BVs infection, are still unknown. In this study, a cDNA library of Bombyx mori cells (BmN) was constructed and yeast two-hybrid screening was used to identify the host cell factors interacting with GP64. One of the eight candidate proteins encoded the E3 ubiquitin-protein ligase SINA-like 10 (SINAL10), was further confirmed through coimmunoprecipitation assays as novel GP64 binding protein. Moreover, overexpression of SINAL10 significantly enhances viral reproduction, and conversely, silencing its expression by small interfering RNAs showed significant inhibitory effects. Collectively, we demonstrated that SINAL10 is a novel GP64-binding protein that stimulates BmNPV proliferation.

Protein-protein interactions of the baculovirus per os infectivity factors (PIFs) in the PIF complex.

After ingestion of occlusion bodies, the occlusion-derived viruses (ODVs) of the baculoviruses establish the first round of infection within the larval host midgut cells. Several ODV envelope proteins, called per os infectivity factors (PIFs), have been shown to be essential for oral infection. Eight PIFs have been identified to date, including P74, PIFs 1-6 and Ac110. At least six PIFs, P74, PIFs 1-4 and PIF6, together with three other ODV-specific proteins, Ac5, P95 (Ac83) and Ac108, have been reported to form a complex on the ODV surface. In this study, in order to understand the interactions of these PIFs, the direct protein-protein interactions of the nine components of the Autographa californica multiple nucleopolyhedrovirus PIF complex were investigated using yeast two-hybrid (Y2H) screening combined with bimolecular fluorescence complementation (BiFC) assay. Six direct interactions, comprising PIF1-PIF2, PIF1-PIF3, PIF1-PIF4, PIF1-P95, PIF2-PIF3 and PIF3-PIF4, were identified in the Y2H analysis, and these results were further verified by BiFC. For P74, PIF6, Ac5 and Ac108, no direct interaction was identified. P95 (Ac83) was identified to interact with PIF1, and further Y2H analysis of the truncation and deletion mutants showed that the predicted P95 chitin-binding domain and amino acids 100-200 of PIF1 were responsible for P95 interaction with PIF1. Furthermore, a summary of the protein-protein interactions of PIFs reported so far, comprising 10 reciprocal interactions and two self-interactions, is presented, which will facilitate our understanding of the characteristics of the PIF complex.

HEp-2 Cell Image Classification With Deep Convolutional Neural Networks.

Efficient Human Epithelial-2 cell image classification can facilitate the diagnosis of many autoimmune diseases. This paper proposes an automatic framework for this classification task, by utilizing the deep convolutional neural networks (CNNs) which have recently attracted intensive attention in visual recognition. In addition to describing the proposed classification framework, this paper elaborates several interesting observations and findings obtained by our investigation. They include the important factors that impact network design and training, the role of rotation-based data augmentation for cell images, the effectiveness of cell image masks for classification, and the adaptability of the CNN-based classification system across different datasets. Extensive experimental study is conducted to verify the above findings and compares the proposed framework with the well-established image classification models in the literature. The results on benchmark datasets demonstrate that 1) the proposed framework can effectively outperform existing models by properly applying data augmentation, 2) our CNN-based framework has excellent adaptability across different datasets, which is highly desirable for cell image classification under varying laboratory settings. Our system is ranked high in the cell image classification competition hosted by ICPR 2014.