Efficient Training on Alzheimer's Disease Diagnosis with Learnable Weighted Pooling for 3D PET Brain Image Classification.

Three-dimensional convolutional neural networks (3D CNNs) have been widely applied to analyze Alzheimer's disease (AD) brain images for a better understanding of the disease progress or predicting the conversion from cognitively impaired (CU) or mild cognitive impairment status. It is well-known that training 3D-CNN is computationally expensive and with the potential of overfitting due to the small sample size available in the medical imaging field. Here we proposed a novel 3D-2D approach by converting a 3D brain image to a 2D fused image using a Learnable Weighted Pooling (LWP) method to improve efficient training and maintain comparable model performance. By the 3D-to-2D conversion, the proposed model can easily forward the fused 2D image through a pre-trained 2D model while achieving better performance over different 3D and 2D baselines. In the implementation, we chose to use ResNet34 for feature extraction as it outperformed other 2D CNN backbones. We further showed that the weights of the slices are location-dependent and the model performance relies on the 3D-to-2D fusion view, with the best outcomes from the coronal view. With the new approach, we were able to reduce 75% of the training time and increase the accuracy to 0.88, compared with conventional 3D CNNs, for classifying amyloid-beta PET imaging from the AD patients from the CU participants using the publicly available Alzheimer's Disease Neuroimaging Initiative dataset. The novel 3D-2D model may have profound implications for timely AD diagnosis in clinical settings in the future.

Inconsistent Performance of Deep Learning Models on Mammogram Classification.

OBJECTIVES: Performance of recently developed deep learning models for image classification surpasses that of radiologists. However, there are questions about model performance consistency and generalization in unseen external data. The purpose of this study is to determine whether the high performance of deep learning on mammograms can be transferred to external data with a different data distribution. MATERIALS AND METHODS: Six deep learning models (three published models with high performance and three models designed by us) were evaluated on four different mammogram data sets, including three public (Digital Database for Screening Mammography, INbreast, and Mammographic Image Analysis Society) and one private data set (UKy). The models were trained and validated on either Digital Database for Screening Mammography alone or a combined data set that included Digital Database for Screening Mammography. The models were then tested on the three external data sets. The area under the receiver operating characteristic curve (auROC) was used to evaluate model performance. RESULTS: The three published models reported validation auROC scores between 0.88 and 0.95 on the validation data set. Our models achieved between 0.71 (95% confidence interval [CI]: 0.70-0.72) and 0.79 (95% CI: 0.78-0.80) auROC on the same validation data set. However, the same evaluation criteria of all six models on the three external test data sets were significantly decreased, only between 0.44 (95% CI: 0.43-0.45) and 0.65 (95% CI: 0.64-0.66). CONCLUSION: Our results demonstrate performance inconsistency across the data sets and models, indicating that the high performance of deep learning models on one data set cannot be readily transferred to unseen external data sets, and these models need further assessment and validation before being applied in clinical practice.

Dynamic Image for 3D MRI Image Alzheimer's Disease Classification.

We propose to apply a 2D CNN architecture to 3D MRI image Alzheimer's disease classification. Training a 3D convolutional neural network (CNN) is time-consuming and computationally expensive. We make use of approximate rank pooling to transform the 3D MRI image volume into a 2D image to use as input to a 2D CNN. We show our proposed CNN model achieves 9.5% better Alzheimer's disease classification accuracy than the baseline 3D models. We also show that our method allows for efficient training, requiring only 20% of the training time compared to 3D CNN models. The code is available online: https://github.com/UkyVision/alzheimer-project.

Heteroduplex DNA in meiotic recombination in Drosophila mei-9 mutants.

Meiotic recombination gives rise to crossovers, which are required in most organisms for the faithful segregation of homologous chromosomes during meiotic cell division. Characterization of crossover-defective mutants has contributed much to our understanding of the molecular mechanism of crossover formation. We report here a molecular analysis of recombination in a Drosophila melanogaster crossover-defective mutant, mei-9. In the absence of mei-9 activity, postmeiotic segregation associated with noncrossovers occurs at the expense of crossover products, suggesting that the underlying meiotic function for MEI-9 is in crossover formation rather than mismatch repair. In support of this, analysis of the arrangement of heteroduplex DNA in the postmeiotic segregation products reveals different patterns from those observed in Drosophila Msh6 mutants, which are mismatch-repair defective. This analysis also provides evidence that the double-strand break repair model applies to meiotic recombination in Drosophila. Our results support a model in which MEI-9 nicks Holliday junctions to generate crossovers during meiotic recombination, and, in the absence of MEI-9 activity, the double Holliday junction intermediate instead undergoes dissolution to generate noncrossover products in which heteroduplex is unrepaired.

REC, Drosophila MCM8, drives formation of meiotic crossovers.

Crossovers ensure the accurate segregation of homologous chromosomes from one another during meiosis. Here, we describe the identity and function of the Drosophila melanogaster gene recombination defective (rec), which is required for most meiotic crossing over. We show that rec encodes a member of the mini-chromosome maintenance (MCM) protein family. Six MCM proteins (MCM2-7) are essential for DNA replication and are found in all eukaryotes. REC is the Drosophila ortholog of the recently identified seventh member of this family, MCM8. Our phylogenetic analysis reveals the existence of yet another family member, MCM9, and shows that MCM8 and MCM9 arose early in eukaryotic evolution, though one or both have been lost in multiple eukaryotic lineages. Drosophila has lost MCM9 but retained MCM8, represented by REC. We used genetic and molecular methods to study the function of REC in meiotic recombination. Epistasis experiments suggest that REC acts after the Rad51 ortholog SPN-A but before the endonuclease MEI-9. Although crossovers are reduced by 95% in rec mutants, the frequency of noncrossover gene conversion is significantly increased. Interestingly, gene conversion tracts in rec mutants are about half the length of tracts in wild-type flies. To account for these phenotypes, we propose that REC facilitates repair synthesis during meiotic recombination. In the absence of REC, synthesis does not proceed far enough to allow formation of an intermediate that can give rise to crossovers, and recombination proceeds via synthesis-dependent strand annealing to generate only noncrossover products.

Detection of Rickettsia amblyommii in association with a tick bite rash.

In the summer of 2006, an Amblyomma americanum tick was removed from a woman in central North Carolina, who subsequently developed a rash at the site of tick attachment. When examined by polymerase chain reaction (PCR) for Borrelia, Anaplasma, Ehrlichia, Babesia, Rickettsia, and Bartonella DNA, only the Rickettsia primers generated an amplicon, which was identified as "R. amblyommii" by sequencing. To our knowledge, this is the first case in which R. amblyommii was temporally associated with a rash.

Prevalence of selected vector-borne organisms and identification of Bartonella species DNA in North American river otters (Lontra canadensis).

Trapper-killed North American river otters (Lontra canadensis) in North Carolina, USA, were screened for multiple vector-borne bacteria known to be pathogenic to mammals. Blood was collected from 30 carcasses in 2006, from 35 in 2007, and from one live otter in 2008. Samples were screened using conventional polymerase chain reaction (PCR) tests for DNA from Bartonella spp., Ehrlichia spp., and spotted fever group Rickettsia spp. All samples were negative for Rickettsia spp. Twelve of 30 samples from 2006 produced amplicons using the assay designed to detect Ehrlichia spp., but sequencing revealed that the amplified DNA fragment was from a novel Wolbachia sp., thought to be an endosymbiote of a Dirofilaria sp. Between 2006 and 2007, DNA from a novel Bartonella sp. was detected in 19 of 65 animals (29%). Blood from one live otter captured in 2008 was found positive for this Bartonella sp. by both PCR and culture. The pathogenicity of this Bartonella species in river otters or other mammals is unknown.