Artificial Intelligence for Breast Ultrasound: AJR Expert Panel Narrative Review.

Breast ultrasound is used in a wide variety of clinical scenarios, including both diagnostic and screening applications. Limitations of ultrasound, however, include its low specificity and, for automated breast ultrasound screening, the time necessary to review whole-breast ultrasound images. As of this writing, four AI tools that are approved or cleared by the FDA address these limitations. Current tools, which are intended to provide decision support for lesion classification and/or detection, have been shown to increase specificity among non-specialists and to decrease interpretation times. Potential future applications include triage of patients with palpable masses in low-resource settings, preoperative prediction of axillary lymph node metastasis, and preoperative prediction of neoadjuvant chemotherapy response. Challenges in the development and clinical deployment of AI for ultrasound include: the limited availability of curated training datasets compared to mammography; the high variability in ultrasound image acquisition due to equipment- and operator-related factors (which may limit algorithm generalizability); and the lack of post-implementation evaluation studies. Furthermore, current AI tools for lesion classification were developed based on 2D data, but diagnostic accuracy could potentially be improved if multimodal ultrasound data were used, such as color Doppler, elastography, cine clips, and 3D imaging.

Standalone AI for Breast Cancer Detection at Screening Digital Mammography and Digital Breast Tomosynthesis: A Systematic Review and Meta-Analysis.

Background There is considerable interest in the potential use of artificial intelligence (AI) systems in mammographic screening. However, it is essential to critically evaluate the performance of AI before it can become a modality used for independent mammographic interpretation. Purpose To evaluate the reported standalone performances of AI for interpretation of digital mammography and digital breast tomosynthesis (DBT). Materials and Methods A systematic search was conducted in PubMed, Google Scholar, Embase (Ovid), and Web of Science databases for studies published from January 2017 to June 2022. Sensitivity, specificity, and area under the receiver operating characteristic curve (AUC) values were reviewed. Study quality was assessed using the Quality Assessment of Diagnostic Accuracy Studies 2 and Comparative (QUADAS-2 and QUADAS-C, respectively). A random effects meta-analysis and meta-regression analysis were performed for overall studies and for different study types (reader studies vs historic cohort studies) and imaging techniques (digital mammography vs DBT). Results In total, 16 studies that include 1 108 328 examinations in 497 091 women were analyzed (six reader studies, seven historic cohort studies on digital mammography, and four studies on DBT). Pooled AUCs were significantly higher for standalone AI than radiologists in the six reader studies on digital mammography (0.87 vs 0.81, P = .002), but not for historic cohort studies (0.89 vs 0.96, P = .152). Four studies on DBT showed significantly higher AUCs in AI compared with radiologists (0.90 vs 0.79, P < .001). Higher sensitivity and lower specificity were seen for standalone AI compared with radiologists. Conclusion Standalone AI for screening digital mammography performed as well as or better than radiologists. Compared with digital mammography, there is an insufficient number of studies to assess the performance of AI systems in the interpretation of DBT screening examinations. © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Scaranelo in this issue.

Comparison of Diagnostic Mammography-Guided Biopsy and Digital Breast Tomosynthesis-Guided Biopsy of Suspicious Breast Calcifications: Results in 1354 Biopsies.

BACKGROUND. Studies have shown improved targeting and sampling of noncalcified lesions (asymmetries, masses, and architectural distortion) with digital breast tomosynthesis (DBT)-guided biopsy in comparison with digital mammography (DM)-guided stereotactic biopsy. Literature that compares the two techniques specifically for sampling calcifications has been scarce. OBJECTIVE. The purpose of this study was to compare the performance and outcomes of DM- and DBT-guided biopsy of suspicious calcifications. METHODS. This retrospective study included 1310 patients (mean age, 58 ± 12 [SD] years) who underwent a total of 1354 9-gauge vacuum-assisted core biopsies of suspicious calcifications performed at a single institution from May 22, 2017, to December 31, 2021. The decision to use a DM-guided or DBT-guided technique was made at the discretion of the radiologist performing the biopsy. Procedure time, the number of exposures during the procedure, and the histopathologic outcomes were recorded. The two techniques were compared using a two-sample t test for continuous variables and a chi-square test for categoric variables. Additional tests were performed using generalized estimating equations to control for the effect of the individual radiologist performing the biopsy. RESULTS. A total of 348 (26%) biopsies used DM guidance, and 1006 (74%) used DBT guidance. The mean procedure time was significantly lower for DBT-guided biopsy (14.9 ± 8.0 [SD] minutes) than for DM-guided biopsy (24.7 ± 14.3 minutes) (p < .001). The mean number of exposures was significantly lower for DBT-guided biopsy (4.1 ± 1.0 [SD] exposures) than for DM-guided biopsy (9.1 ± 3.3 exposures) (p < .001). The differences in procedure time and number of exposures remained significant (both p < .001) when controlling for the effect of the radiologist performing the biopsy. There were no significant differences (all p > .05) between DM-guided and DBT-guided biopsy in terms of the malignancy rate on initial biopsy (20% vs 19%), the rate of high-risk lesion upgrading (14% vs 22%), or the final malignancy rate (23% vs 22%). CONCLUSION. DBT-guided biopsy of suspicious calcifications can be performed with shorter procedure time and fewer exposures compared with DM-guided biopsy, without a significant difference in rates of malignancy or high-risk lesion upgrading. CLINICAL IMPACT. The use of a DBT-guided, rather than a DM-guided, biopsy technique for suspicious calcifications can potentially reduce patient discomfort and radiation exposure without affecting clinical outcomes.

High-risk lesions in the breast diagnosed by MRI-guided core biopsy: upgrade rates and features associated with malignancy.

PURPOSE: This study assessed the upgrade rates of high-risk lesions (HRLs) in the breast diagnosed by MRI-guided core biopsy and evaluated imaging and clinical features associated with upgrade to malignancy. METHODS: This IRB-approved, retrospective study included MRI-guided breast biopsy exams yielding HRLs from August 1, 2011, to August 31, 2020. HRLs included atypical ductal hyperplasia (ADH), lobular carcinoma in situ (LCIS), atypical lobular hyperplasia (ALH), radial scar, and papilloma. Only lesions that underwent excision or at least 2 years of MRI imaging follow-up were included. For each HRL, patient history, imaging features, and outcomes were recorded. RESULTS: Seventy-two lesions in 65 patients were included in the study, with 8/72 (11.1%) of the lesions upgraded to malignancy. Upgrade rates were 16.7% (2/12) for ADH, 100% (1/1) for pleomorphic LCIS, 40% (2/5) for other LCIS, 0% (0/19) for ALH, 0% (0/18) for papilloma, and 0% (0/7) for radial scar/complex sclerosing lesion. Additionally, two cases of marked ADH bordering on DCIS and one case of marked ALH bordering on LCIS, were upgraded. Lesions were more likely to be upgraded if they presented as T2 hypointense (versus isotense, OR 6.46, 95% CI 1.27-32.92) or as linear or segmental non-mass enhancement (NME, versus focal or regional, p = 0.008). CONCLUSION: Our data support the recommendation that ADH and LCIS on MRI-guided biopsy warrant surgical excision due to high upgrade rates. HRLs that present as T2 hypointense, or as linear or segmental NME, should be viewed with suspicion as these were associated with higher upgrade rates to malignancy.

Developing Asymmetries without Sonographic Correlate at Digital Breast Tomosynthesis.

Background Digital breast tomosynthesis (DBT) has improved the accuracy of mammography, including resolving many breast asymmetries as overlapping breast tissue. The pathologic outcomes of persistent developing asymmetries visualized at DBT are not well established. Purpose To characterize the outcomes and the predictors of malignancy for developing asymmetries visualized at DBT without a sonographic correlate. Materials and Methods This retrospective study included all tomosynthesis-guided biopsies of developing asymmetries performed at a single institution from May 2017 through January 2020. A reader study including three breast imaging radiologists determined interrater agreement and inclusion into the study. Electronic medical records were used to extract patient characteristics, imaging characteristics, and pathologic diagnoses. The Wilcoxon rank sum test, Fisher exact test, and χ2 test were used to analyze correlations of patient and imaging characteristics with likelihood of malignancy. Results The reader study included 95 DBT examinations with moderate interrater reliability (Fleiss κ = 0.45). There was majority reader agreement in 85 of the 95 DBT examinations (89%) of 83 women (median age, 56 years; interquartile range, 47-69 years), and this finalized the study data set. At pathologic examination, most asymmetries (68 of 85, 80%) were benign, with common diagnoses being fibrocystic change (n = 20), stromal fibrosis (n = 10), and fat necrosis (n = 10). The overall malignancy rate was 20% (17 of 85 asymmetries; 95% CI: 12, 29); 15 of the 17 malignancies (88%) were invasive cancers. Malignancies were more common in women with a personal history of breast cancer (35% vs 10%, P = .02). Conclusion In 85 developing asymmetries visualized at digital breast tomosynthesis without a sonographic correlate, there was a 20% (95% CI: 12, 29) malignancy rate, which was higher than the rates of malignancy for a developing asymmetry detected at digital mammography. © RSNA, 2021 See also the editorial by Skaane in this issue.

Artificial Intelligence for Reducing Workload in Breast Cancer Screening with Digital Breast Tomosynthesis.

Background Digital breast tomosynthesis (DBT) has higher diagnostic accuracy than digital mammography, but interpretation time is substantially longer. Artificial intelligence (AI) could improve reading efficiency. Purpose To evaluate the use of AI to reduce workload by filtering out normal DBT screens. Materials and Methods The retrospective study included 13 306 DBT examinations from 9919 women performed between June 2013 and November 2018 from two health care networks. The cohort was split into training, validation, and test sets (3948, 1661, and 4310 women, respectively). A workflow was simulated in which the AI model classified cancer-free examinations that could be dismissed from the screening worklist and used the original radiologists' interpretations on the rest of the worklist examinations. The AI system was also evaluated with a reader study of five breast radiologists reading the DBT mammograms of 205 women. The area under the receiver operating characteristic curve (AUC), sensitivity, specificity, and recall rate were evaluated in both studies. Statistics were computed across 10 000 bootstrap samples to assess 95% CIs, noninferiority, and superiority tests. Results The model was tested on 4310 screened women (mean age, 60 years ± 11 [standard deviation]; 5182 DBT examinations). Compared with the radiologists' performance (417 of 459 detected cancers [90.8%], 477 recalls in 5182 examinations [9.2%]), the use of AI to automatically filter out cases would result in 39.6% less workload, noninferior sensitivity (413 of 459 detected cancers; 90.0%; P = .002), and 25% lower recall rate (358 recalls in 5182 examinations; 6.9%; P = .002). In the reader study, AUC was higher in the standalone AI compared with the mean reader (0.84 vs 0.81; P = .002). Conclusion The artificial intelligence model was able to identify normal digital breast tomosynthesis screening examinations, which decreased the number of examinations that required radiologist interpretation in a simulated clinical workflow. Published under a CC BY 4.0 license. Online supplemental material is available for this article. See also the editorial by Philpotts in this issue.

COVID-19 Vaccine-Related Axillary Adenopathy on Breast Imaging: Follow-Up Recommendations and Histopathologic Findings.

This study describes 94 patients who presented with suspected COVID-19 vaccine-related axillary adenopathy on breast imaging. All biopsies recommended within 12 weeks of the second vaccine dose were benign. Among women not recommended for biopsy, the median interval between the second vaccine dose and ultrasound follow-up was 15.9 weeks. Three biopsies yielding malignant diagnoses were recommended 12.0-13.1 weeks after the second vaccine dose. Lengthening imaging follow-up to 12-16 weeks after the second dose may reduce unnecessary biopsy recommendations.

Comparison of Outcomes for One-View Asymmetries Recalled From Digital Breast Tomosynthesis Versus Full-Field Digital Mammography Screening Examinations.

BACKGROUND. Recall rates are lower for digital breast tomosynthesis (DBT) than for full-field digital mammography (FFDM). This difference could have important implications with respect to one-view asymmetries given that missed cancers are often visible on one view. OBJECTIVE. The purpose of this study is to compare the outcomes of one-view asymmetries recalled from DBT versus FFDM screening examinations and to determine predictors of malignancy among recalled asymmetries. METHODS. This retrospective study first determined recall rates associated with one-view asymmetries for screening mammography performed using DBT and FFDM from July 14, 2016, through July 14, 2020. Further analyses included patients recalled for a one-view asymmetry who completed subsequent diagnostic workup and all recommended follow-up. Patient and cancer characteristics were extracted from the electronic health record. RESULTS. The recall rate associated with asymmetries was lower for DBT screening (2.5% [3169/128,755]) than for FFDM screening (3.4% [815/23,898]) (p < .001). Further analyses of patients who completed diagnostic workup and subsequent follow-up included 3119 patients (mean age, 57 years) for DBT screening and 811 patients (mean age, 56 years) for FFDM screening. Distribution of final BI-RADS categories from subsequent diagnostic workup was not different between the two modalities (p > .99). The frequency of malignancy was not different between asymmetries recalled from DBT (1.7% [54/3119]) and FFDM (1.7% [14/811]) (p > .99). Malignant asymmetries identified on FFDM versus DBT were more frequently associated with architectural distortion on diagnostic workup (35.7% [5/14] vs 9.3% [5/54]) (p < .001) and were more commonly invasive ductal carcinoma (92.9% vs 57.4%) and less commonly invasive lobular carcinoma (0.0% vs 24.1%) (p = .05). In multivariable analysis, independent predictors of malignancy among recalled asymmetries from DBT were age (for 55-69 years, odds ratio [OR] = 2.40 [p = .04]; for ≥ 70 years, OR = 7.93 [p < .001]; reference, < 55 years) and breast density (not dense, OR = 2.47 [p = .001]; reference, dense breasts). CONCLUSION. Recalled asymmetries were less frequent for DBT than for FFDM. The malignancy rate was low for both modalities (1.7%). Age 55 years old and older and lower breast density predicted malignancy for DBT-recalled asymmetries. CLINICAL IMPACT. Our results support the use of DBT to reduce unnecessary recalls without altering PPV for asymmetry-associated malignancies. Patient factors should be considered when assessing whether a potential asymmetry on DBT screening represents overlapping fibroglandular tissue or a suspicious finding that requires diagnostic workup.

Comparison of emergency pediatric breast ultrasound interpretations and management recommendations between pediatric radiologists and breast imaging radiologists.

PURPOSE: Pediatric patients with breast-related symptoms often initially present to the emergency department for evaluation. While pediatric radiologists are accustomed to evaluating acute infectious and traumatic etiologies, they may be less familiar with breast-specific findings. This study compares management recommendations of pediatric breast ultrasounds performed in the emergency setting between pediatric and breast imaging radiologists. METHODS: This retrospective cohort study reviewed data from all pediatric breast ultrasounds performed in the emergency setting from a single academic institution from 1/1/14 to 12/31/19. During the study period, 12 pediatric radiologists with experience ranging from 1 to 33 years interpreted pediatric breast ultrasounds. Three breast imaging radiologists (with 3, 8, and 25 years of experience) retrospectively reviewed each case and recorded whether further management was recommended. Differences in recommendations were compared using Fisher's exact test. Cohen's kappa was used to assess agreement between subspecialty radiologists. RESULTS: This study included 75 pediatric patients, with mean age 13 ± 5.6 years and malignancy rate of 1.3% (1/75). Pediatric radiologists and the most experienced breast imaging radiologist had moderate agreement in management recommendations (k = 0.54). There was no significant difference in recommendations for further management between pediatric radiologists (22/75 [29.3%]) and the most experienced breast imaging radiologist (15/75 [20.0%]), p = 0.26. CONCLUSION: Recommendations for pediatric breast complaints in the emergency setting are comparable between subspecialties.

Diagnostic Versus Screening Mammography Recommendations for Postlumpectomy Imaging Surveillance of Patients With Breast Cancer.

Among 707 women who were recommended to undergo annual diagnostic mammography (DM) surveillance after lumpectomy for breast cancer, 94.9%, 90.4%, and 84.3% of women presented for DM at years 1, 2, and 3 after lumpectomy. A total of 18.8%, 11.0%, and 9.9% of women received additional views at years 1, 2, and 3, compared with the 10.1% institutional screening recall rate. The postlumpectomy year 3 cancer detection rate of 11.7 cancers per 1000 DM examinations was below DM benchmarks. These preliminary findings suggest that returning to screening mammography may be acceptable after 1 year of postlumpectomy DM follow-up.

Tomosynthesis-Guided Vacuum-Assisted Breast Biopsy of Architectural Distortion Without a Sonographic Correlate: A Retrospective Review.

BACKGROUND. Digital breast tomosynthesis-guided vacuum-assisted breast biopsy (DBT VAB) allows biopsy of findings seen better or exclusively on digital breast tomosynthesis (DBT), including architectural distortion. Although architectural distortion with an associated sonographic mass correlate has a high risk of malignancy, limited data describe the radiologic-pathologic correlation of tomosynthesis-detected architectural distortion without a sonographic correlate. OBJECTIVE. This study evaluates the malignancy rate of architectural distortions without a sonographic correlate that undergo DBT VAB and provides radiologic-pathologic correlation for benign, high-risk, and malignant entities that are associated with architectural distortion. METHODS. We retrospectively reviewed imaging, as well as pathology slides and/or reports, for DBT VABs performed for architectural distortion without a sonographic correlate at a single institution between June 1, 2017, and January 15, 2020. According to the correlative histopathology, cases were categorized as benign, high risk, or malignant, and specific histopathologic diagnoses were summarized. RESULTS. During the study period, 142 patients (mean age, 59 years) underwent DBT VAB for 151 unique architectural distortions without a sonographic correlate. DBT VAB revealed a malignant diagnosis in 27 (18%), a high-risk lesion in 50 (33%), and a benign diagnosis in 74 (49%). Two cases of atypical ductal hyperplasia were upgraded to malignancy, resulting in a final malignancy rate of 19% (n = 29/151). Most malignant lesions were invasive carcinomas (83%, n = 24/29); most invasive carcinomas were of lobular subtype (54%, n = 13/24). Most high-risk lesions were radial scars/complex sclerosing lesions (62%, n = 31/50). Most benign results represented fibrocystic change (66%, n = 49/74). A subset (11%, n = 8/74) of benign results were considered discordant and subsequently excised, with none representing malignancy. CONCLUSION. The final malignancy rate of 19% in architectural distortion without a sonographic correlate justifies a recommendation for biopsy using DBT VAB. CLINICAL IMPACT. Our results highlight the utility of DBT VAB in the era of DBT. The detailed radiologic-pathologic correlations will assist radiologists in assessing concordance when performing DBT VAB for architectural distortions and provide a reference for future patient management.

DeepCAT: Deep Computer-Aided Triage of Screening Mammography.

Although much deep learning research has focused on mammographic detection of breast cancer, relatively little attention has been paid to mammography triage for radiologist review. The purpose of this study was to develop and test DeepCAT, a deep learning system for mammography triage based on suspicion of cancer. Specifically, we evaluate DeepCAT's ability to provide two augmentations to radiologists: (1) discarding images unlikely to have cancer from radiologist review and (2) prioritization of images likely to contain cancer. We used 1878 2D-mammographic images (CC & MLO) from the Digital Database for Screening Mammography to develop DeepCAT, a deep learning triage system composed of 2 components: (1) mammogram classifier cascade and (2) mass detector, which are combined to generate an overall priority score. This priority score is used to order images for radiologist review. Of 595 testing images, DeepCAT recommended low priority for 315 images (53%), of which none contained a malignant mass. In evaluation of prioritizing images according to likelihood of containing cancer, DeepCAT's study ordering required an average of 26 adjacent swaps to obtain perfect review order. Our results suggest that DeepCAT could substantially increase efficiency for breast imagers and effectively triage review of mammograms with malignant masses.

Outcomes of Foci on Breast MRI: Features Associated With Malignancy.

OBJECTIVE. The objective of this study was to determine the outcomes of foci seen on breast MRI and to evaluate imaging features associated with malignancy. MATERIALS AND METHODS. In this institutional review board-approved retrospective study, we reviewed 200 eligible foci in 179 patients that were assigned BI-RADS category of 3 or 4 from December 2004 to August 2018. Clinical and imaging features of all eligible foci were collected, and associations with malignant outcomes were evaluated. Malignancy rates were also calculated. RESULTS. Of 200 eligible foci, 64 were assigned BI-RADS category 3 and 136 were assigned BI-RADS category 4. The malignancy rate was 1.6% (1/64) among BI-RADS 3 foci and 17.6% (24/136) for BI-RADS 4 foci. The majority of malignant foci represented invasive breast cancer (68.0%, 17/25). Focus size and washout kinetics were significantly associated with malignant outcome (p < 0.05). CONCLUSION. Despite the high prevalence of foci on breast MRI, data are limited to guide their management. Foci should not be disregarded, because foci undergoing biopsy had a malignancy rate of 17.6%, with the majority of malignant foci representing invasive cancer. Larger size and washout kinetics were associated with malignancy in our study and should raise the suspicion level for a focus on breast MRI.

Breast MRI biopsy cancellation due to lesion nonvisualization.

It is a clinical dilemma when a finding reported as suspicious on a breast MRI is not visualized at the time of a scheduled MRI-guided breast biopsy. We retrospectively reviewed all canceled MRI-guided biopsies at our institution between 6/1/2009 and 9/20/2019 and found a cancellation rate of 6.9% (72/1051). In one case, a mastectomy was performed after the canceled biopsy revealing a focus of DCIS in the same quadrant as the original finding (malignancy rate 2.1%). Our results support the current practice of 6-month follow-up MRI recommendation after a canceled MRI-guided biopsy for lesion nonvisualization.

Engineering of TIMP-3 as a LAP-fusion protein for targeting to sites of inflammation.

Tissue inhibitor of metalloproteinase (TIMP)-3 is a natural inhibitor of a range of enzymes that degrade connective tissue and are involved in the pathogenesis of conditions such as arthritis and cancer. We describe here the engineering of TIMP-3 using a novel drug-delivery system known as the 'LAP technology'. This involves creating therapeutic proteins in fusion with the latency-associated peptide (LAP) from the cytokine TGF-? to generate proteins that are biologically inactive until cleavage of the LAP to release the therapy. LAP-TIMP-3 was successfully expressed in mammalian cells and the presence of the LAP resulted in a 14-fold increase in the quantity of recombinant TIMP-3 produced. LAP-TIMP-3 was latent until release from the LAP by treatment with matrix metalloproteinase when it could inhibit proteases of the adamalysins and adamalysins with thrombospondin motifs families, but not matrix metalloproteinases, indicating that this version of TIMP-3 is a more specific inhibitor than the native protein. There was sufficient protease activity in synovial fluid from human joints with osteoarthritis to release TIMP-3 from the LAP fusion. These results demonstrate the potential for development of TIMP-3 as a novel therapy for conditions where upregulation of catabolic enzymes are part of the pathology.

A Pilot of Data-Driven Modeling to Assess Potential for Improved Efficiency in an Academic Breast-Imaging Center.

Patient satisfaction and department efficiency are central pillars in defining quality in medicine. Patient satisfaction is often linked to wait times. We describe a novel method to study workflow and simulate solutions to improve efficiency, thereby decreasing wait times and adding value. We implemented a real-time location system (RTLS) in our academic breast-imaging department to study workflow, including measuring patient wait time, quantifying equipment utilization, and identifying bottlenecks. Then, using discrete event simulation (DES), we modeled solutions with changes in staffing and equipment. Nine hundred and ninety-nine patient encounters were tracked over a 10-week period. The RTLS system recorded 551,512 raw staff and patient time stamps, which were analyzed to produce 17,042 staff and/or patient encounter time stamps. Mean patient wait time was 27 min. The digital breast tomosynthesis (DBT) unit had the highest utilization rate and was identified as a bottleneck. DES predicts a 19.2% reduction in patient length of stay with replacement of a full field digital mammogram (FFDM) unit by a DBT unit and the addition of technologists. Through integration of RTLS with discrete event simulation testing, we created a model based on real-time data to accurately assess patient wait times and patient progress through an appointment, evaluate patient staff-interaction, identify system bottlenecks, and quantitate potential solutions. This quality improvement initiative has important implications, potentially allowing data-driven decisions for staff hiring, equipment purchases, and department layout.

'The few cubic centimetres inside your skull': a neurological reading of George Orwell's Nineteen Eighty-Four.

Nineteen Eighty-Four (1949), George Orwell's political satire on state surveillance and mind control, was written between 1946 and 1948, at a time when new thinking in forensic psychiatry coincided with scientific breakthroughs in neurology to bring questions of criminality, psychotherapy and mental health to the forefront of the popular imagination. This paper examines how Nineteen Eighty-Four inverts psychiatric paradigms in order to diagnose what Orwell sees as the madness of totalitarian regimes. It then goes on to place the novel's dystopian vision of total surveillance and mind control in the context of the neurological research and brain scanning techniques of the mid-20th century. Not only does this context provide new insight into the enduring power of Orwell's novel, it also locates it within a historical moment when technological interventions into the brain seemed to offer a paradigm of mental health and illness as a simple, knowable binary. Nineteen Eighty-Four complicates this binary, and deserves to be acknowledged as an early example of what might be called 'electric shock' literature, within a mid-20th century canon that includes Harold Pinter's The Caretaker (1960), Ken Kesey's One Flew Over the Cuckoo's Nest (1962), and Sylvia Plath's The Bell Jar (1963).