Mask-Guided Convolutional Neural Network for Breast Tumor Prognostic Outcome Prediction on 3D DCE-MR Images

Journal of Digital Imaging - In this proof-of-concept work, we have developed a 3D-CNN architecture that is guided by the tumor mask for classifying several patient-outcomes in breast cancer from...     

Endoplasmic reticulum stress is involved in hydrogen peroxide induced apoptosis in immortalized and malignant human oral keratinocytes

Background: Although hydrogen peroxide may play an important role in the development of cancer, it can be an efficient inducer of apoptosis in cancer cells; the exact mechanism by which this action occurs is not completely understood in oral cancer cells. Method: In this study, the mechanisms by which H₂O₂ inhibited growth and induced apoptosis were differentially investigated using HPV‐immortalized human oral keratinocytes (IHOK) and oral cancer cells (HN4). Results: H₂O₂ treatment sensitively and dose‐dependently induced growth inhibition and typical apoptosis in IHOK and HN4 cells, as demonstrated by a decreased level of cell viability, an increased population of cells in the sub‐G₀/G₁ phase, ladder formation of the genomic DNA, chromatin condensation and accumulation of Annexin V⁺/PI⁺ cells. Furthermore, the expression of Bax, p53 and p21^WAF1/CIP1 increased, whereas the expression of Bcl‐2 decreased in immortalized and malignant keratinocytes that were treated with H₂O₂. In addition, cytochrome‐c from the mitochondria was observed in H₂O₂‐treated IHOK and oral cancer cells, and this was accompanied by the activation of caspase‐3 and ‐9. Additionally, H₂O₂ treatment induced upregulation of CHOP, GRP78 and several representative endoplasmic reticulum (ER) stress‐responsive proteins, including heme oxygenase‐1. Conclusion: Overall, these results suggest that H₂O₂ triggers apoptosis via the mitochondrial and ER stress pathway in IHOK and HN4 cells, and that increasing the cellular levels of H₂O₂ sufficiently may lead to selective killing of oral cancer cells and therefore be therapeutically useful.     

The effect of internalizing human single chain antibody fragment on liposome targeting to epithelioid and sarcomatoid mesothelioma

Immunoliposomes (ILs) anchored with internalizing human antibodies capable of targeting all subtypes of mesothelioma can be useful for targeted imaging and therapy of this malignant disease. The objectives of this study were to evaluate both the in vitro and in vivo tumor targeted internalization of novel internalizing human single chain antibody (scFv) anchored ILs on both epithelioid (M28) and sarcomatoid (VAMT-1) subtypes of human mesothelioma. ILs were prepared by post-insertion of mesothelioma-targeting human scFv (M1) onto preformed liposomes and radiolabeled with (111)In ((111)In-IL-M1), along with control non-targeted liposomes ((111)In-CL). Incubation of (111)In-IL-M1 with M28, VAMT-1, and a control non-tumorigenic cell line (BPH-1) at 37 °C for 24 h revealed efficient binding and rapid internalization of ILs into both subtypes of tumor cells but not into the BPH-1 cells; internalization accounted for approximately 81-94% of total cell accumulation in mesothelioma cells compared to 37-55% in control cells. In tumor-bearing mice intravenous (i.v.) injection of (111)In-IL-M1 led to remarkable tumor accumulation: 4% and 4.7% injected dose per gram (% ID/g) for M28 and VAMT-1 tumors, respectively, 48 h after injection. Furthermore, tumor uptake of (111)In-IL-M1 in live xenograft animal models was verified by single photon emission computed tomography (SPECT/CT). In contrast, i.v. injection of (111)In-CL in tumor-bearing mice revealed very low uptake in both subtypes of mesothelioma, 48 h after injection. In conclusion, M1 scFv-anchored ILs showed selective tumor targeting and rapid internalization into both epithelioid and sarcomatoid subtypes of human mesothelioma, demonstrating its potential as a promising vector for enhanced tumor drug targeting.     

Partial-volume correction in PET: validation of an iterative postreconstruction method with phantom and patient data.

Partial-volume errors (PVEs) in PET can cause incorrect estimation of radiopharmaceutical uptake in small tumors. An iterative postreconstruction method was evaluated that corrects for PVEs without a priori knowledge of tumor size or background. METHODS: Volumes of interest (VOIs) were drawn on uncorrected PET images. PVE-corrected images were produced using an iterative 3-dimensional deconvolution algorithm and a local point spread function. The VOIs were projected on the corrected image to estimate the PVE-corrected mean activity concentration. These corrected mean values were compared with uncorrected maximum and mean values. Simulated data were generated as a first test of the correction algorithm. Phantom measurements were made using (18)F-FDG-filled spheres in a scattering medium. Clinical validation used 154 surrogate tumors from 9 patients. The surrogate tumors were blood-pool images of the descending aorta as well as mesenteric and iliac arteries and veins. Surrogate tumors ranged in diameter from 5 to 25 mm. Analysis used (18)F-FDG and (11)C-CO datasets (both dynamic and static). Values representing "truth" were derived from imaging the blood pool in large structures (e.g., the left ventricle, left atrium, or sections of the aorta) where PVEs were negligible. Surrogate tumor sizes were measured from contrast CT. RESULTS: The PVE-correction technique, when applied to the mean value in spheric phantoms, yielded recovery coefficients of 87% for an 8-mm-diameter sphere and between 100% and 103% for spheres between 13 and 29 mm. For the human studies, PVE-corrected data recovered a large fraction of the true activity concentration (86% +/- 7% for an 8-mm-diameter tumor and 98% +/- 8% for tumors between 10 and 24 mm). For tumors smaller than 18 mm, the PVE-corrected mean values were less biased (P<0.05) than the uncorrected maximum or mean values. CONCLUSION: Iterative postreconstruction PVE correction generated more accurate uptake measurements in subcentimeter tumors for both phantoms and patients than the uncorrected values. The method eliminates the requirement for segmenting anatomic data and estimating tumor metabolic size or tumor background level. This technique applies a PVE correction to the mean voxel value within a VOI, yielding a more accurate estimate of uptake than the maximum voxel value.     

Progressive hair loss and myocardial degeneration in rough coat mice: reduced lysyl oxidase-like (LOXL) in the skin and heart

The rough coat (rc) is a spontaneous recessive mutation in mice. To identify the mutated gene, we have characterized the rc phenotype and initiated linkage mapping. The rc mice show growth retardation, cyclic and progressive hair loss, hyperplastic epidermis, abnormal hair follicles, cardiac muscle degeneration, and reduced amount of collagen and elastin in the skin and heart. The rc locus was mapped at 32.0 cM on chromosome 9, close to the loxl gene. Lysyl oxidase-like (LOXL) protein is a novel copper-containing amine oxidase that is required for the cross-linking of elastin and collagen in vitro. LOXL is expressed at high levels in the skin and heart, where the rc mice show strong phenotype. The expression pattern and the genetic proximity to rc suggested loxl as a potential candidate gene. In rc mice, the loxl mRNA was reduced in the skin and the LOXL protein in the heart, dermis, atrophic hair follicles, and sebaceous glands. No mutations, however, were identified within the coding region of loxl, and offspring from rc/rc and loxl null mice crossing were phenotypically normal. Based on these results, loxl appears non-allelic to rc. Heart- and skin-specific downregulation of LOXL in rc mice, however, may contribute to the extracellular matrix alterations and the rc phenotype.     

How to make a machine think in art psychotherapy: An expert system's reasoning process

In making decisions, a human being considers all of the factors concerning the situation in a comprehensive and intuitive manner by using all of his or her experience and knowledge. Due to the inherent nature of human decision making, the reasoning process from confronting the problem to finding a solution is too complicated to be explicitly represented. In this paper, we implement an expert system for the diagnosis process of art psychotherapists. We model the complicated mechanism of this process as several procedural stages and feedbacks. We devise a suitable method of maintaining consistency among numerous decisions derived from the system. We also provide the system with a learning facility to improve its intelligence. Finally, we demonstrate the usefulness and suitability of the proposed system through a case study.