Deep Learning for Radiotherapy Outcome Prediction Using Dose Data - A Review.

Artificial intelligence, and in particular deep learning using convolutional neural networks, has been used extensively for image classification and segmentation, including on medical images for diagnosis and prognosis prediction. Use in radiotherapy prognostic modelling is still limited, however, especially as applied to toxicity and tumour response prediction from radiation dose distributions. We review and summarise studies that applied deep learning to radiotherapy dose data, in particular studies that utilised full three-dimensional dose distributions. Ten papers have reported on deep learning models for outcome prediction utilising spatial dose information, whereas four studies used reduced dimensionality (dose volume histogram) information for prediction. Many of these studies suffer from the same issues that plagued early normal tissue complication probability modelling, including small, single-institutional patient cohorts, lack of external validation, poor data and model reporting, use of late toxicity data without taking time-to-event into account, and nearly exclusive focus on clinician-reported complications. They demonstrate, however, how radiation dose, imaging and clinical data may be technically integrated in convolutional neural networks-based models; and some studies explore how deep learning may help better understand spatial variation in radiosensitivity. In general, there are a number of issues specific to the intersection of radiotherapy outcome modelling and deep learning, for example translation of model developments into treatment plan optimisation, which will require further combined effort from the radiation oncology and artificial intelligence communities.

Airflow characteristics in a baboon nasal passage cast.

Airflow patterns in the nasal passages influence the distribution of air-pollutant-induced lesions in the airway mucosa. Little is known about airflow characteristics of the complex nasopharyngeal airway of humans and experimental animals. Airflow characteristics in the nasopharyngeal airways of an adult male baboon (13.9 kg body wt) were investigated with thermistor probes and the findings compared with flow visualization, using a cinephotographic technique. A clear, acrylic, hollow cast of a baboon nose was made, and thermistor probes were inserted to record air velocity in the cast lumen using a wind tunnel to propel air through the cast. An identical cast was studied by passing water through the cast, with pulses of dye to reveal flow, and cinephotography was used for determination of flow velocities and flow patterns. Flow rates adjusted on the basis of a Reynolds conversion showed good correlations between the two methods, whereas cinephotography revealed areas of turbulence and vortex-like flow not detected by thermistor probes. These results suggest that water flow may provide useful information in complex airways where airflow cannot be determined by other methods.

A morphometric comparison of the nasopharyngeal airway of laboratory animals and humans.

Solid silicone rubber casts of the nasopharyngeal and laryngeal regions of a human cadaver (child, 3 years old) and a laboratory primate (baboon, 10 years old) were made, and cross-sectional areas were measured in detail. Cross-sectional areas of other species reported in the published literature were used for comparison. In the child's nose cast, the frontal nasal duct (frontonasal duct), which enters the anterior part of the middle meatus, and the sphenoidal recess were almost absent. The ethmoidal turbinates (superior and middle concha) and the maxillary turbinates (inferior concha) were present but were not fully developed. In the baboon nose, the different turbinates were well defined and smooth but of a less complex nature than the child's nose. Of the species compared, the baboon's upper airways had the greatest similarity to the human child's. The present study shows that for the species investigated and for those from the literature, the cross-sectional area increases from the external nares to the maxilloturbinate region (inferior concha). There is a relatively sudden drop in cross-sectional area about halfway through the nose. The present study suggests a functional relationship between nasal structure and cross-sectional area across species.

A morphometric study of nasal-pharyngeal growth for particle deposition in the rat.

Animal studies frequently are used in assessing potential human health effects from exposure to inhaled toxicants. Such studies also are used to investigate sensitive subpopulations such as children. Among other factors that influence the degree to which animal models are predictive of human effects in the delivered dose of the toxicant to the various regions of the respiratory tract. Because the rat is an obligatory nose breather, an understanding of the rat nasal-pharyngeal airway geometry is needed to relate exposures to delivered doses. In this study, the growth and development of the rat nasal-pharyngeal airway was studied at one-week intervals in male Fischer-344 rats from one to five weeks. Casts of an adult (60 day) and an aging (441 day) rat were included for comparison. Replica casts of the nasal-pharyngeal airway were made by injecting silicone rubber through the trachea, and sections in anterior-posterior positions were made for morphometric study. A simple structure of the nasal-pharyngeal airway was found in the young rats. While the percentage of the airway composed of turbinates was similar at all ages, the surface area of the turbinates increased 7.7-fold between 7 and 60 days. Because of the simpler structure and smaller surface area in the young rat, extrathoracic clearance is probably less efficient, resulting in a higher delivered dose to the lung of a young rat than to that of an adult rat exposed to the same toxicant concentration.