Histology segmentation using active learning on regions of interest in oral cavity squamous cell carcinoma.

In digital pathology, deep learning has been shown to have a wide range of applications, from cancer grading to segmenting structures like glomeruli. One of the main hurdles for digital pathology to be truly effective is the size of the dataset needed for generalization to address the spectrum of possible morphologies. Small datasets limit classifiers' ability to generalize. Yet, when we move to larger datasets of whole slide images (WSIs) of tissue, these datasets may cause network bottlenecks as each WSI at its original magnification can be upwards of 100 000 by 100 000 pixels, and over a gigabyte in file size. Compounding this problem, high quality pathologist annotations are difficult to obtain, as the volume of necessary annotations to create a classifier that can generalize would be extremely costly in terms of pathologist-hours. In this work, we use Active Learning (AL), a process for iterative interactive training, to create a modified U-net classifier on the region of interest (ROI) scale. We then compare this to Random Learning (RL), where images for addition to the dataset for retraining are randomly selected. Our hypothesis is that AL shows benefits for generating segmentation results versus randomly selecting images to annotate. We show that after 3 iterations, that AL, with an average Dice coefficient of 0.461, outperforms RL, with an average Dice Coefficient of 0.375, by 0.086.

Primary intestinal-type adenocarcinoma of the oral tongue: Case report and review of histologic origin and oncologic management.

BACKGROUND: Primary intestinal-type adenocarcinoma (ITAC) of the oral tongue is an extremely uncommon malignancy with only 3 cases reported in the literature. This high-grade malignancy originates from metaplasia of minor salivary glands. METHODS: A 40-year-old man presented with a gradually enlarging midline oral tongue mass, odynophagia, and dysphagia. Management included a median lingual glossectomy, bilateral neck dissections, and adjuvant chemoradiation with 5-fluorouracil (5-FU). Additional chemotherapy with folinic acid, fluorouracil, and oxaliplatin (FOLFOX) was given to mimic therapy in colonic adenocarcinomas. RESULTS: Thirteen months after surgery and adjuvant chemoradiotherapy, there is no evidence of locoregional or distant disease. His diet and speech have normalized after reconstruction without free tissue transfer. CONCLUSION: We report the fourth case of oral tongue ITAC, and present the first histologic evidence of metaplasia of oral cavity salivary epithelium. We also discuss adjuvant therapy recommendations given the lack of clarity for treatment of this rare disease.

Inverse planning of energy-modulated electron beams in radiotherapy.

The use of megavoltage electron beams often poses a clinical challenge in that the planning target volume (PTV) is anterior to other radiosensitive structures and has variable depth. To ensure that skin as well as the deepest extent of the PTV receives the prescribed dose entails prescribing to a point beyond the depth of peak dose for a single electron energy. This causes dose inhomogeneities and heightened potential for tissue fibrosis, scarring, and possible soft tissue necrosis. Use of bolus on the skin improves the entrant dose at the cost of decreasing the therapeutic depth that can be treated. Selection of a higher energy to improve dose homogeneity results in increased dose to structures beyond the PTV, as well as enlargement of the volume receiving heightened dose. Measured electron data from a linear accelerator was used as input to create an inverse planning tool employing energy and intensity modulation using bolus (e-IMRT). Using tools readily available in a radiotherapy department, the applications of energy and intensity modulation on the central axis makes it possible to remove hot spots of 115% or more over the depths clinically encountered. The e-IMRT algorithm enables the development of patient-specific dose distributions with user-defined positions of peak dose, range, and reduced dose to points beyond the prescription point.

Monoallelic deletion of the p53 gene through chromosomal translocation in a small cell osteosarcoma.

Small cell osteosarcoma is a rare bone tumor of high-grade malignancy that most often arises in the metaphysis of long bones in the second decade of life. Cytogenetic and molecular genetic findings in small cell osteosarcoma are poorly defined. Conventional cytogenetic analysis of a small cell osteosarcoma arising in the proximal tibia of a 9-year-old male revealed a diploid chromosomal complement with complex structural rearrangements involving chromosomes 6, 16, and 17. Immunohistochemical assessment of p53 protein expression revealed nuclear p53 immunoreactivity in approximately 15% of the neoplastic cells. Subsequent fluorescence in situ hybridization (FISH) analyses confirmed loss of the p53 gene locus on the derivative chromosome 17 homolog and were negative for amplification of the MDM2, CDK4, c-MYC, HER-2/neu, CCND1, and COPS3 gene loci. To the best of our knowledge, this represents the first demonstration of monoallelic deletion of p53 in small cell osteosarcoma, suggesting that p53 alterations may play an important role in the development of small cell osteosarcoma as well as conventional osteosarcoma.

Consequences of multiple withdrawals from alcohol.

This article represents the proceedings of a symposium at the 2003 annual meeting of the Research Society on Alcoholism in Fort Lauderdale, FL, organized by Theodora Duka and chaired by Dai Stephens. The purpose of the symposium was to examine the effects of multiple experiences of withdrawal from alcohol in animals made dependent on alcohol and in humans who are alcohol dependent. Parallels were drawn to the effects of repeated short-lived high-content alcohol exposures in animals and in humans who are social drinkers but indulge in binge drinking. The presentations were (1) Multiple detoxifications and risk of relapse in abstinent alcoholics, by John Gentry and Robert Malcolm; (2) Emotional and cognitive impairments after long-term use of alcohol: relationship to multiple detoxifications and binge drinking, by Theodora Duka; (3) The effect of repeated withdrawal from ethanol on conditioning to appetitive stimuli, by Tamzin Ripley, Gilyanna Borlikova, and Dai Stephens; (4) Alcohol withdrawal kindling: electrographic measures in a murine model of behavioral seizure sensitization, by Lynn Veatch and Howard Becker; and (5) Binge drinking induced changes in CNS, by Fulton Crews.

New anticonvulsants: a review of applications for the management of substance abuse disorders.

Substance abuse and related disorders are a major problem in the United States and elsewhere. In many cases, pharmacotherapy has been shown to be useful in the treatment of substance abuse and related disorders, both in detoxification and in the prevention of relapse. However, currently used medications are not ideal, due either to their effectiveness being less than desirable or their side effect profiles being greater than desirable, or both. Medications with increased effectiveness or decreased side effects are needed. This article reviews three anticonvulsant medications, gabapentin, oxcarbazepine, and tiagabine, for possible utility in the treatment of substance abuse and related disorders. The following aspects of each medication are discussed: metabolism and kinetics, dosage and administration, possible interactions, side effects, and applications to substance abuse treatment; the last subject includes discussion of management of withdrawal symptoms, treatment of dual diagnosis, management of substance abuse sleep disorders, direct suppression of substance abuse/relapse prevention, and abuse potential. Each medication appears to have potential advantages for these uses, and further study of these medications in treatment of substance abuse disorders is recommended.