Active, continual fine tuning of convolutional neural networks for reducing annotation efforts.

The splendid success of convolutional neural networks (CNNs) in computer vision is largely attributable to the availability of massive annotated datasets, such as ImageNet and Places. However, in medical imaging, it is challenging to create such large annotated datasets, as annotating medical images is not only tedious, laborious, and time consuming, but it also demands costly, specialty-oriented skills, which are not easily accessible. To dramatically reduce annotation cost, this paper presents a novel method to naturally integrate active learning and transfer learning (fine-tuning) into a single framework, which starts directly with a pre-trained CNN to seek "worthy" samples for annotation and gradually enhances the (fine-tuned) CNN via continual fine-tuning. We have evaluated our method using three distinct medical imaging applications, demonstrating that it can reduce annotation efforts by at least half compared with random selection.

Computer-aided detection and visualization of pulmonary embolism using a novel, compact, and discriminative image representation.

Diagnosing pulmonary embolism (PE) and excluding disorders that may clinically and radiologically simulate PE poses a challenging task for both human and machine perception. In this paper, we propose a novel vessel-oriented image representation (VOIR) that can improve the machine perception of PE through a consistent, compact, and discriminative image representation, and can also improve radiologists' diagnostic capabilities for PE assessment by serving as the backbone of an effective PE visualization system. Specifically, our image representation can be used to train more effective convolutional neural networks for distinguishing PE from PE mimics, and also allows radiologists to inspect the vessel lumen from multiple perspectives, so that they can report filling defects (PE), if any, with confidence. Our image representation offers four advantages: (1) Efficiency and compactness-concisely summarizing the 3D contextual information around an embolus in only three image channels, (2) consistency-automatically aligning the embolus in the 3-channel images according to the orientation of the affected vessel, (3) expandability-naturally supporting data augmentation for training CNNs, and (4) multi-view visualization-maximally revealing filling defects. To evaluate the effectiveness of VOIR for PE diagnosis, we use 121 CTPA datasets with a total of 326 emboli. We first compare VOIR with two other compact alternatives using six CNN architectures of varying depths and under varying amounts of labeled training data. Our experiments demonstrate that VOIR enables faster training of a higher-performing model compared to the other compact representations, even in the absence of deep architectures and large labeled training sets. Our experiments comparing VOIR with the 3D image representation further demonstrate that the 2D CNN trained with VOIR achieves a significant performance gain over the 3D CNNs. Our robustness analyses also show that the suggested PE CAD is robust to the choice of CT scanner machines and the physical size of crops used for training. Finally, our PE CAD is ranked second at the PE challenge in the category of 0 mm localization error.

Convolutional Neural Networks for Medical Image Analysis: Full Training or Fine Tuning?

Training a deep convolutional neural network (CNN) from scratch is difficult because it requires a large amount of labeled training data and a great deal of expertise to ensure proper convergence. A promising alternative is to fine-tune a CNN that has been pre-trained using, for instance, a large set of labeled natural images. However, the substantial differences between natural and medical images may advise against such knowledge transfer. In this paper, we seek to answer the following central question in the context of medical image analysis: Can the use of pre-trained deep CNNs with sufficient fine-tuning eliminate the need for training a deep CNN from scratch? To address this question, we considered four distinct medical imaging applications in three specialties (radiology, cardiology, and gastroenterology) involving classification, detection, and segmentation from three different imaging modalities, and investigated how the performance of deep CNNs trained from scratch compared with the pre-trained CNNs fine-tuned in a layer-wise manner. Our experiments consistently demonstrated that 1) the use of a pre-trained CNN with adequate fine-tuning outperformed or, in the worst case, performed as well as a CNN trained from scratch; 2) fine-tuned CNNs were more robust to the size of training sets than CNNs trained from scratch; 3) neither shallow tuning nor deep tuning was the optimal choice for a particular application; and 4) our layer-wise fine-tuning scheme could offer a practical way to reach the best performance for the application at hand based on the amount of available data.

siRNA-mediated knock-down of COX-2 in melanocytes suppresses melanogenesis.

Cyclooxygenase-2 (COX-2) is an enzyme induced in response to multiple mitogenic and inflammatory stimuli, including UV light. UV-induced COX-2 expression induces production of prostaglandin E2 (PGE2) in keratinocytes, which mediates inflammation and cell proliferation. Until recently, studies regarding COX-2 and PGE2 in the skin have focused on keratinocytes and skin cancer and the effect of PGs produced by keratinocytes on melanocytes. However, the effects of COX-2 itself or COX-2 inhibitors on melanogenesis are not well known. Therefore, to establish the role of COX-2 in melanogenesis, we investigated the effects of knock-down of COX-2 in melanocytes on melanin production and the expression of melanogenic molecules through silencing of COX-2 expression with COX-2 short interfering RNA (siRNA). COX-2 knock-down in melanocytes decreased the expressions of tyrosinase, TRP-1, TRP-2, gp100 and MITF and also reduced tyrosinase enzyme activity. Furthermore, COX-2 siRNA-transfected melanocytes showed markedly reduced alpha-melanocyte stimulating hormone (α-MSH)-induced melanin production. In addition, α-MSH-induced COX-2 expression in both scrambled siRNA-transfected and COX-2 siRNA-transfected melanocytes was greater than α-MSH-untreated cells. Our results suggest that COX-2 might be a candidate target for the development of anti-melanogenic agents and α-MSH-induced pigmentation could be closely associated with COX-2 expression. COX-2 inhibitors might therefore be of particular use in whitening cosmetics for hyperpigmentation disorders such as melasma, postinflammatory hyperpigmentation and solar lentigo.

A search for synthetic peptides that inhibit soluble N-ethylmaleimide sensitive-factor attachment receptor-mediated membrane fusion.

Soluble N-ethylmaleimide sensitive-factor attachment receptor (SNARE) proteins have crucial roles in driving exocytic membrane fusion. Molecular recognition between vesicle-associated (v)-SNARE and target membrane (t)-SNARE leads to the formation of a four-helix bundle, which facilitates the merging of two apposing membranes. Synthetic peptides patterned after the SNARE motifs are predicted to block SNARE complex formation by competing with the parental SNAREs, inhibiting neuronal exocytosis. As an initial attempt to identify the peptide sequences that block SNARE assembly and membrane fusion, we created thirteen 17-residue synthetic peptides derived from the SNARE motifs of v- and t-SNAREs. The effects of these peptides on SNARE-mediated membrane fusion were investigated using an in vitro lipid-mixing assay, in vivo neurotransmitter release and SNARE complex formation assays in PC12 cells. Peptides derived from the N-terminal region of SNARE motifs had significant inhibitory effects on neuroexocytosis, whereas middle- and C-terminal-mimicking peptides did not exhibit much inhibitory function. N-terminal mimicking peptides blocked N-terminal zippering of SNAREs, a rate-limiting step in SNARE-driven membrane fusion. Therefore, the results suggest that the N-terminal regions of SNARE motifs are excellent targets for the development of drugs to block SNARE-mediated membrane fusion and neurotransmitter release.