Performance Metric Analysis of a Radiation Oncology Educational Podcast

PURPOSE/OBJECTIVE(S)

Asynchronous podcast education is a popular supplementary tool with up to 88% of medical residents reporting its use, and is perceived by faculty to have high educational value with convenience and connection to broader medical communities.1,2 Radiation oncology (RO) podcasts remain scarce compared to other specialties, and ones focused exclusively on education are largely absent. We analyze the early performance, listenership, and engagement of the first RO medical education podcast.

MATERIALS/METHODS

Episode data and listener demographics were gathered from Spotify and iTunes. Episode engagement was defined as a percentage of plays on unique devices playing >40% of an episode within a single session. Listenership was defined by the number of plays per day (ppd) on unique devices, averaged over 60 days from publication date. Episodes were case based, categorized by disease subsite, and reviewed by a board-certified radiation oncologist. Quantitative endpoints included episode engagement and listenership. Qualitative comments were not solicited but received through email and Twitter. Pearson's correlation coefficient calculations were used for analysis.

RESULTS

Eighteen total episodes had 8,517 total plays since July 2022 over 176 days. Median episode length was 13.8 minutes (range 9.2-20.1). Popular listening platforms included iTunes (53.5%) and Spotify (34.0%). Listener demographics included 59.4% male, 39.6% female, and 1.0% other, ranging from age 23-27 (14%), 28-34 (65%), 35-44 (14%), 45-59 (4%), and 60+ (1%). ATB was played in 48 countries, with the most listeners in North America (74.6%) followed by Asia (7.8%), Europe (7.6%), Australia (7.0%), Africa (2.0%), and South America (0.4%). There was a 464% increase in listenership since publication with median growth of 63.3% per month. Median listenership and engagement were 9.2 ppd (IQR, 7.7-9.9) and 77.8% (IQR, 68.1-81.2) for all episodes, respectively. Among 8 topics, head and neck (HN) episodes had the highest mean listenership with 17.8 ppd, followed by genitourinary (GU, 10.8) and lung (10.5). GU episodes had the highest mean engagement at 84.6%, followed by lung (82.3) and sarcoma (81.2). Dosimetry had the lowest listenership and engagement at 5.9 ppd and 63.1%, respectively. A significant negative relationship between episode length and engagement was observed, (r(18) = -0.469, p = 0.05). There was no statistically significant relationship between ppd and episode length, (r(18) = -0.303, p = 0.22).

CONCLUSION

Evidenced by its significant rise in listenership, high listener engagement, and large international audience, there were previously unmet needs for RO medical education that may be supplemented by podcasts. HN episodes were most popular with GU exhibiting highest engagement. Longer episode length correlated with a significant decrease in engagement but no effect on popularity.

Reinforcement Learning Powered Station Parameter Optimized Radiation Therapy (SPORT): A Novel Treatment Planning and Beam Delivery Technique

PURPOSE/OBJECTIVE(S)

Conventional intensity modulated radiation therapy (IMRT) with a typical 5-20 fixed beams often does not provide sufficient angular sampling required for conformal dose shaping, whereas current volumetric modulated arc therapy (VMAT) discretizes the angular space into equally spaced control points without considering the differential need for intensity modulation of different angles, leading to undersampling at some angles while oversampling at some other angles. Our goal is to develop a node or station parameter optimized radiation therapy (SPORT) strategy with simultaneously optimized angular sampling and beam modulation by leveraging state-of-the-art reinforcement learning and the unique capability of modern digital LINACs in dose delivery through programmable nodal points.

MATERIALS/METHODS

We developed a SPORT optimization framework, in which, the process of programming control points (or station parameters) was formulated as a stochastic dynamic programming problem, which was solved by a reinforcement learning-based algorithm. On-policy reinforcement learning method, namely, state-action-reward-state-action (SARSA) was integrated with deep convolutional neural network to predict station parameters by utilizing the patient's anatomical structures meanwhile considering the delivery capability of a typical digital LINAC machine. Here, the deep convolutional neural network estimated the state-action value by using the quality of the plan with current station parameters when a next potential station parameter was selected. The state-action value was then updated by SARSA learning. The quality of the plan was quantified by dosimetry constraints. The model was assessed by a retrospective study on a cohort of patients underwent head-and-neck radiation therapy. Dosimetric analysis and delivery efficiency comparisons were used to evaluate the performance of the proposed framework.

RESULTS

Our model was used to generate 16 plans unseen in the original training set. All the plans predicted by our model achieved better dose distributions without violating clinical planning constraints. Moreover, instead of using 4 full standard arcs in the original clinically used plans obtained via manual optimization, the predicted plans only used one full standard arc (about 178 control points) plus boost from a few sub-arcs (less than 30 degrees of gantry angles), which significantly improved the efficiency of the beam delivery. We are in the process of integrating the sub-arcs into the full arc by considering the programmable capability of modern LINACs.

CONCLUSION

We demonstrated that a machine learning-based SPORT framework capable of optimizing the spatial sampling and beam modulation simultaneously for modern radiation therapy. The framework not only significantly improves the quality and efficiency of beam delivery, but also has the potential to be incorporated into current clinical workflow to improve the efficiency of dose planning and delivery.

Clinical Implementation of an Automated IMRT/VMAT Treatment Planning Tool

PURPOSE/OBJECTIVE(S)

To create an in-house automated treatment planning tool for IMRT/VMAT treatments and evaluate the dosimetric plan quality against manually generated plans.

MATERIALS/METHODS

A scripting application programming interface is employed to interact with a commercial treatment planning system (TPS) to implement automatic plan evaluation and update optimization parameters by mimicking the human planning process. The automated planning performs in an iterative fashion until reaching an acceptable tradeoff among target coverage/dose homogeneity and sparing of critical organs at risk. In each iteration, the dose constraints, priorities, and optimization structures for are automatically updated based on the results of the current iteration. Twenty previously treated plans (10 prostate and 10 head and neck), were preliminarily used to evaluate the performance of the automated planning tool. The differences in target and organ-at-risk metrics from the manually generated clinical plans were analyzed using paired t-test to evaluate clinical acceptability of tour automated planning tool. The current in-house-developed automated planning solution is able to create plans for different disease sites, including head & neck, prostate, pelvis, and lung. So far, the VMAT plans for more than 150 different cases have been generated with the tool. The results for these were also evaluated.

RESULTS

Compared to the manually generated clinical head and neck plans, all auto plans achieved PTV D95% coverage and critical organs at risk sparing without statistically significant change in average global Dmax (107.4% for manual vs 107.3% for automated plans). The auto-planning solution provided reduced maximum doses to brainstem and spinal cord (average reductions with standard deviations of 5.1 ± 2.6 Gy and 2.9 ± 1.4 Gy, respectively, all p <0.03), reduced average mean doses to contralateral parotid, ipsilateral parotid, contralateral submandibular gland, pharynx, esophagus, cochleae (reductions of 2.2 ± 2.9 Gy, 4.8 ± 4.7 Gy, 3.6 ± 5.2 Gy, 2.0 ± 7.1 Gy, 3.9 ± 2.6 Gy, 3.8 ± 5.0 Gy, respectively, all p < 0.045). Similar results were observed for the prostate plans. With the same PTV coverage and without statistically significant change in average global Dmax (106.5% for manual vs 106.8% for automated plans), the automated solution provided superior sparing for both bladder and rectum. Bladder V75, V70, V65 were reduced by 0.6% ± 2.1%, 0.8% ± 2.5%, and 0.9% ± 2.9% (all p <0.04), respectively. Rectum V75, V70, V65, V60 were reduced by 1.0% ± 2.3%, 1.2% ± 2.8%, 1.3% ± 3.2%, 1.6% ± 3.6% (all p < 0.01), respectively.

CONCLUSION

Our automated treatment planning solution is capable of efficiently generating VMAT plans for different disease sites with superior dosimetric indices compared to manually generated plans. Our tool is integrated within a commercial TPS platform, so it has the advantage of seamless adoption into the standard workflow to improve plan quality and treatment planning efficiency in our clinic.

Prospective Clinical Adoption of Artificial Intelligence for Organ Contouring in Head and Neck Radiation Treatment Planning

PURPOSE/OBJECTIVE(S)

Patients that undergo head and neck (H&N) radiation therapy (RT) require laborious delineation of organs-at-risk (OARs) on computed tomography (CT) scans in a treatment planning system (TPS) to minimize radiation to normal tissue. This task can be completed rapidly and accurately with recently developed artificial intelligence-based semantic segmentation models. The current study aims to deploy and evaluate a strategy for improving clinical practice with this technology.

MATERIALS/METHODS

Deep learning models were trained and tested with CT scans and OAR contours from previous H&N RT cases at our clinic. Two medical physicists vetted the models and selected a 2.5D U-Net for further implementation. The model was embedded in a dedicated server at the hospital, programmed to read H&N CT scans staged for import into the TPS, generate auto-contours, and write them into a TPS-compatible format made available alongside the scan. In the pilot implementation, the auto-contouring service was utilized for more than 60 cases, prospectively. The auto-contours were quantitatively evaluated against the treatment-approved contours to determine how much modification was performed by the clinical team.

RESULTS

The 2.5D U-Net selected for clinical integration segments 21 OARs in less than 3 minutes per scan. Across all the prospective cases, the mean Dice score and mean 95th percentile Hausdorff distance (mm) between the auto-contour and treatment-approved contour for each of the 21 OARs were as follows, respectively: brainstem (0.93, 1.94), optic chiasm (0.70, 2.96), left cochlea (0.69, 2.37), right cochlea (0.68, 2.44), esophagus (0.88, 2.46), left globe (0.93, 1.50), right globe (0.93, 1.63), glottis (0.91, 2.13), larynx (0.93, 2.76), mandible (0.90, 4.86), left optic nerve (0.78, 1.64), right optic nerve (0.82, 1.65), oral cavity (0.86, 8.46), left parotid gland (0.91, 2.78), right parotid gland (0.91, 2.39), pharynx (0.85, 2.39), spinal cord (0.87, 2.27), left submandibular gland (0.85, 3.46), right submandibular gland (0.83, 3.69), left temporal lobe (0.94, 2.20), and right temporal lobe (0.95, 2.09). The auto-contours for the optic chiasm, optic nerves, cochleas, and submandibular glands differed substantially from the final contours, a finding corroborated by the clinical team; the rest were clinically acceptable with minor or no edits necessary.

CONCLUSION

The proposed strategy provides a sophisticated starting point for treatment planning that has garnered overall favorable feedback from the participating radiation oncologists and dosimetrists. Consequently, the technique is being extended to other treatment sites.

Interaction energies between beta-lactam antibiotics and E. coli penicillin-binding protein 5 by reversible thermal denaturation

Penicillin-binding proteins (PBPs) catalyze the final stages of bacterial cell wall biosynthesis. PBPs form stable covalent complexes with beta-lactam antibiotics, leading to PBP inactivation and ultimately cell death. To understand more clearly how PBPs recognize beta-lactam antibiotics, it is important to know their energies of interaction. Because beta-lactam antibiotics bind covalently to PBPs, these energies are difficult to measure through binding equilibria. However, the noncovalent interaction energies between beta-lactam antibiotics and a PBP can be determined through reversible denaturation of enzyme-antibiotic complexes. Escherichia coli PBP 5, a D-alanine carboxypeptidase, was reversibly denatured by temperature in an apparently two-state manner with a temperature of melting (T(m)) of 48.5 degrees C and a van't Hoff enthalpy of unfolding (H(VH)) of 193 kcal/mole. The binding of the beta-lactam antibiotics cefoxitin, cloxacillin, moxalactam, and imipenem all stabilized the enzyme significantly, with T(m) values as high as +4.6 degrees C (a noncovalent interaction energy of +2.7 kcal/mole). Interestingly, the noncovalent interaction energies of these ligands did not correlate with their second-order acylation rate constants (k(2)/K'). These rate constants indicate the potency of a covalent inhibitor, but they appear to have little to do with interactions within covalent complexes, which is the state of the enzyme often used for structure-based inhibitor design.

Inhibition of AmpC beta-lactamase through a destabilizing interaction in the active site

Beta-lactamases hydrolyze beta-lactam antibiotics, including penicillins and cephalosporins; these enzymes are the most widespread resistance mechanism to these drugs and pose a growing threat to public health. beta-Lactams that contain a bulky 6(7)alpha substituent, such as imipenem and moxalactam, actually inhibit serine beta-lactamases and are widely used for this reason. Although mutant serine beta-lactamases have arisen that hydrolyze beta-lactamase resistant beta-lactams (e.g., ceftazidime) or avoid mechanism-based inhibitors (e.g., clavulanate), mutant serine beta-lactamases have not yet arisen in the clinic with imipenemase or moxalactamase activity. Structural and thermodynamic studies suggest that the 6(7)alpha substituents of these inhibitors form destabilizing contacts within the covalent adduct with the conserved Asn152 in class C beta-lactamases (Asn132 in class A beta-lactamases). This unfavorable interaction may be crucial to inhibition. To test this destabilization hypothesis, we replaced Asn152 with Ala in the class C beta-lactamase AmpC from Escherichia coli and examined the mutant enzyme's thermodynamic stability in complex with imipenem and moxalactam. Consistent with the hypothesis, the Asn152 --> Ala substitution relieved 0.44 and 1.10 kcal/mol of strain introduced by imipenem and moxalactam, respectively, relative to the wild-type complexes. However, the kinetic efficiency of AmpC N152A was reduced by 6300-fold relative to that of the wild-type enzyme. To further investigate the inhibitor's interaction with the mutant enzyme, the X-ray crystal structure of moxalactam in complex with N152A was determined to a resolution of 1.83 A. Moxalactam in the mutant complex is significantly displaced from its orientation in the wild-type complex; however, moxalactam does not adopt an orientation that would restore competence for hydrolysis. Although Asn152 forces beta-lactams with 6(7)alpha substituents out of a catalytically competent configuration, making them inhibitors, the residue is essential for orienting beta-lactam substrates and cannot simply be replaced with a much smaller residue to restore catalytic activity. Designing beta-lactam inhibitors that interact unfavorably with this conserved residue when in the covalent adduct merits further investigation.

Functional analyses of AmpC beta-lactamase through differential stability

Despite decades of intense study, the complementarity of beta-lactams for beta-lactamases and penicillin binding proteins is poorly understood. For most of these enzymes, beta-lactam binding involves rapid formation of a covalent intermediate. This makes measuring the equilibrium between bound and free beta-lactam difficult, effectively precluding measurement of the interaction energy between the ligand and the enzyme. Here, we explore the energetic complementarity of beta-lactams for the beta-lactamase AmpC through reversible denaturation of adducts of the enzyme with beta-lactams. AmpC from Escherichia coli was reversibly denatured by temperature in a two-state manner with a temperature of melting (Tm) of 54.6 degrees C and a van't Hoff enthalpy of unfolding (deltaH(VH)) of 182 kcal/mol. Solvent denaturation gave a Gibbs free energy of unfolding in the absence of denaturant (deltaG(u)H2O) of 14.0 kcal/mol. Ligand binding perturbed the stability of the enzyme. The penicillin cloxacillin stabilized AmpC by 3.2 kcal/mol (deltaTm = +5.8 degrees C); the monobactam aztreonam stabilized the enzyme by 2.7 kcal/mol (deltaTm = +4.9 degrees C). Both acylating inhibitors complement the active site. Surprisingly, the oxacephem moxalactam and the carbapenem imipenem both destabilized AmpC, by 1.8 kcal/mol (deltaTm = -3.2 degrees C) and 0.7 kcal/mol (deltaTm = -1.2 degrees C), respectively. These beta-lactams, which share nonhydrogen substituents in the 6(7)alpha position of the beta-lactam ring, make unfavorable noncovalent interactions with the enzyme. Complexes of AmpC with transition state analog inhibitors were also reversibly denatured; both benzo(b)thiophene-2-boronic acid (BZBTH2B) and p-nitrophenyl phenylphosphonate (PNPP) stabilized AmpC. Finally, a catalytically inactive mutant of AmpC, Y150F, was reversibly denatured. It was 0.7 kcal/mol (deltaTm = -1.3 degrees C) less stable than wild-type (WT) by thermal denaturation. Both the cloxacillin and the moxalactam adducts with Y150F were significantly destabilized relative to their WT counterparts, suggesting that this residue plays a role in recognizing the acylated intermediate of the beta-lactamase reaction. Reversible denaturation allows for energetic analyses of the complementarity of AmpC for beta-lactams, through ligand binding, and for itself, through residue substitution. Reversible denaturation may be a useful way to study ligand complementarity to other beta-lactam binding proteins as well.

Comparing the thermodynamic stabilities of a related thermophilic and mesophilic enzyme

Several models have been proposed to explain the high temperatures required to denature enzymes from thermophilic organisms; some involve greater maximum thermodynamic stability for the thermophile, and others do not. To test these models, we reversibly melted two analogous protein domains in a two-state manner. E2cd is the isolated catalytic domain of cellulase E2 from the thermophile Thermomonospora fusca. CenAP30 is the analogous domain of the cellulase CenA from the mesophile Cellulomonas fimi. When reversibly denatured in a common buffer, the thermophilic enzyme E2cd had a temperature of melting (Tm) of 72.2 degrees C, a van't Hoff enthalpy of unfolding (DeltaHVH) of 190 kcal/mol, and an entropy of unfolding (DeltaSu) of 0.55 kcal/(mol*K); the mesophilic enzyme CenAP30 had a Tm of 56.4 degrees C, a DeltaHVH of 107 kcal/mol, and a DeltaSu of 0. 32 kcal/(mol*K). The higher DeltaHVH and DeltaSu values for E2cd suggest that its free energy of unfolding (DeltaGu) has a steeper dependence on temperature at the Tm than CenAP30. This result supports models that predict a greater maximum thermodynamic stability for thermophilic enzymes than for their mesophilic counterparts. This was further explored by urea denaturation. Under reducing conditions at 30 degrees C, E2cd had a concentration of melting (Cm) of 5.2 M and a DeltaGu of 11.2 kcal/mol; CenAP30 had a Cm of 2.6 M and a DeltaGu of 4.3 kcal/mol. Under nonreducing conditions, the Cm and DeltaGu of CenAP30 were increased to 4.5 M and 10.8 kcal/mol at 30 degrees C; the Cm for E2cd was increased to at least 7.4 M at 32 degrees C. We were unable to determine a DeltaGu value for E2cd under nonreducing conditions due to problems with reversibility. These data suggest that E2cd attains its greater thermal stability (DeltaTm = 15.8 degrees C) through a greater thermodynamic stability (DeltaDeltaGu = 6.9 kcal/mol) compared to its mesophilic analogue CenAP30.

Longitudinal vibrations of a silica fiber segment characterized using a fiber Bragg grating

A modulator assembly that excites longitudinal vibrations in a short (19 cm) silica fiber segment is described. A physical model of the modulator assembly is used to theoretically predict the behavior of longitudinal vibrations in the fiber segment. The longitudinal vibrations are experimentally characterized using an intrinsic fiber Bragg grating. Experimental results are compared to theoretical predictions.