Pith. sign in

REVIEW 3 major objections 5 minor 39 references

Mapping Order in Semicrystalline Polymers using Machine Learning of Nanobeam Electron Diffraction

T0 review · 3 major / 5 minor · reviewed 2026-08-01 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A U-Net trained on 500,000 synthetic electron diffraction patterns detects lamellar, backbone, and π-π peaks in noisy polymer 4DSTEM data more accurately and about 5.5x faster than correlative template matching, with one user-set threshold.

desk verdict A genuine ML advance for polymer 4DSTEM peak detection, with solid synthetic validation, but the experimental head-to-head is undercut by a simulator tuned on the same sample and no independent ground truth -- still worth a serious referee. read the letter →

arxiv 2607.16570 v1 pith:B3QNNIIS submitted 2026-07-18 cond-mat.mtrl-sci cs.LG

Nicholas Marchese , Arthur R. C. McCray , Yael Tsarfati , Karen Bustillo , Adam Marks , Alberto Salleo , Colin Ophus This is my paper · ORCID
classification cond-mat.mtrl-scics.LG PACS 61.05.jm68.37.Lp
keywords semicrystallinepolymers4DSTEMnanobeamelectrondiffractionU-Netpeakdetectionsynthetictrainingdataorganicmixedionicelectronicconductorsmachinelearning
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper tackles a bottleneck in 4D scanning transmission electron microscopy of semicrystalline polymers: the three kinds of diffraction peaks (lamellar stacking, backbone periodicity, π-π stacking) look very different and are buried in low-dose noise, so template-matching algorithms miss many of them. The authors train a U-Net on half a million synthetic diffraction patterns that mimic beam-sensitive polymer data, teaching it to output both peak positions and intensities. On an experimental oxidized p(g3T2) sample, the network finds peaks in 99.65% of probe positions versus 85.16% for the correlative method, and runs about 5.5 times faster. If the synthetic-to-real transfer holds, this makes near-live mapping of polymer microstructure during microscope acquisition feasible and removes most operator-tuned parameters.

What carries the argument

The load-bearing piece is the synthetic diffraction-pattern generator, which creates 500,000 training examples by specifying recipes of components — central beam, amorphous halo, lamellar, backbone, and π-π peaks with varied intensities, radial positions, annular spreads, asymmetries, and Poisson noise — plus exact label images for peak positions and intensities. The U-Net is trained to regress those two label channels; at inference, a Gaussian filter, local-maximum criterion, and subpixel quadratic fit extract the final peaks, and the intensity channel is sampled at those coordinates. The single user threshold sets the detection cutoff on the position channel.

What would settle it

Record the same polymer area at low dose and at high dose (fresh cryo region), identify peaks unambiguously in the high-dose data, and check the model's low-dose detections against that ground truth; if agreement is no better than template matching's, the synthetic-to-real transfer has failed. Alternatively, feed the model synthetic patterns from an independent generator with different noise and peak statistics and measure the F1 drop.

Watch

Extended reading notes

Core claim

The central claim is that a U-Net trained purely on synthetic data can detect polymer diffraction peaks in real noisy 4DSTEM datasets more accurately and faster than conventional correlative template matching. The model outputs two maps per diffraction pattern: a peak-position channel and a peak-intensity channel, from which peak coordinates and intensities are extracted with standard local-maximum detection. Across the three polymer peak morphologies, the network detects peaks in 31,288 of 31,397 probe positions (99.65%), compared to 26,737 (85.16%) for the correlative algorithm, and processes the dataset in 3 min 20 s versus 18 min 21 s. The authors also show that orientation maps built fr

Load-bearing premise

The synthetic diffraction-pattern generator, tuned initially to one oxidized p(g3T2) dataset and then broadened, produces training examples that match real experimental polymer diffraction well enough that the model's synthetic validation performance transfers to the microscope.

Editorial extensions

If this is right

  • Near-live visualization: each 256×256 scan position can be processed in a fraction of the acquisition time, so operators could see orientation maps while still at the microscope.
  • More complete orientation/order maps: lamellar domains that appear patchy under template matching are revealed as nearly continuous, changing how apparent disorder is interpreted.
  • The same model transfers to other polymer systems (reduced p(g3T2), reduced PB2T-TEG) without retraining, suggesting broader applicability to weakly scattering molecular or oxide materials.
  • Removing operator-tuned hyperparameters (15 for template matching down to 1 threshold) reduces user bias and makes analysis reproducible across labs.
  • The validation establishes dose thresholds (~60–80 counts for lamellar/backbone, 150–800 for π-π) below which even this model cannot detect peaks, framing detectability limits.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Because the experimental comparison's verification is visual, a decisive next test is a high-dose ground-truth dataset; this is an editorial extension, not a paper claim.
  • The model's generality is bounded by the original p(g3T2) training seed; running it on deliberately different polymer chemistries, detector geometries, or dose regimes would map how far the synthetic-to-real transfer extends.
  • The same synthetic-labeling approach could be extended to output peak class labels or orientation fields directly, and reused for other beam-sensitive materials where labelled experimental data are impossible to obtain.
  • Near-live processing implies operators could close the loop during acquisition; the paper notes this but leaves adaptive experiments to future work.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The manuscript trains a U-Net on 500,000 synthetic nanobeam electron diffraction patterns to detect three classes of semicrystalline polymer reflections (lamellar, backbone, and π-π) in noisy 4DSTEM data. The model outputs peak positions and intensities, with peaks extracted by local-maximum detection on the position channel and intensities sampled from the intensity channel. The authors validate the model on held-out synthetic data using F1 scores as a function of peak intensity, radial distance, and annular spread. They then compare the ML model with correlative template matching on an experimental oxidized p(g3T2) dataset, reporting coverage counts (99.65% vs. 85.16% of probe positions with at least one peak detected), orientation maps, and a 5.5× speedup. Generality is demonstrated on two additional polymer systems, reduced p(g3T2) and reduced PB2T-TEG. The central claim is that the ML model is faster and outperforms correlative algorithms in almost all cases.

Significance. If the experimental accuracy claim were established, this would be a useful practical tool for 4DSTEM analysis of beam-sensitive semicrystalline polymers: the synthetic-training approach avoids manual labeling, the F1 curves provide quantitative detection thresholds, and the reported speed would enable near-real-time mapping. The manuscript is also candid about several limitations, including visual inspection as the only experimental ground truth and known edge cases. However, the central experimental accuracy claim is not yet established because the synthetic generator was tuned on the same oxidized p(g3T2) sample used as the benchmark and because the experimental comparison lacks independent ground truth. The synthetic validation is thorough in scope, but it does not by itself transfer the accuracy claim to real experimental data.

major comments (3)
  1. [§2 (Methods) and §3 (Results, Figs. 4–5)] The synthetic generator's parameters were 'initially tuned with reference to experimental data from an oxidized p(g3T2) dataset,' and the primary experimental benchmark is the same oxidized p(g3T2) sample. The model is not trained on experimental labels, so this is not full circularity, but the benchmark cannot establish that the simulator is representative: good agreement could reflect tuning to this material. The experimental comparison has no independent ground truth—'inspected by eye' by the authors with surrounding-probe context. Please validate on an independent material not used in generator tuning and/or use blind expert labels or synthetic peaks injected into experimental patterns.
  2. [§3, 'Experimental results' (Fig. 4)] The headline numbers (99.65% vs. 85.16% of probe positions with at least one peak) are coverage counts, not precision/recall. A detector that emits more false positives also achieves higher coverage. The difference maps in Fig. 4 categorize detections as 'ML only,' 'Corr only,' etc., but the categorization was done by visual inspection by the authors with context; no quantitative false-positive rate is given. Please report precision/recall against an independent label set, or at minimum a blinded expert-labeling study, before claiming ML 'outperforms correlative algorithms in almost all cases.'
  3. [§3, processing-time comparison] The processing-time comparison states '3 minutes and 20 seconds compared to ... 18 minutes and 21 seconds, translating to ... 5.5× faster ... when run on a GPU.' It is not stated whether the correlative algorithm was run on the same GPU, with comparable I/O and optimized code. The speed advantage and the near-real-time claim in §4 depend on this. Please specify hardware, library versions, and whether both methods used the same device; otherwise report the comparison as not hardware-matched.
minor comments (5)
  1. [§2, Methods] Typo: 'V oronoi' should be 'Voronoi.'
  2. [§2, Eq. (1)] Define all symbols in Eq. (1) explicitly in the main text (d_e, w_bkg, I_p, I_T). The term 'peak counts' is ambiguous: it refers to expected electron counts per peak, not the number of detected peaks.
  3. [§3, Fig. 3] The F1 curves do not include error bars or confidence intervals. Since 1,000 synthetic patterns are evaluated per condition, bootstrap confidence intervals would help substantiate the sharp thresholds and the high-intensity degradation observed for π-π peaks.
  4. [§3, synthetic validation] The validation set contains only isolated peaks of each type, whereas the training set includes multi-peak patterns with orientational context. Reporting F1 on multi-peak validation patterns would better reflect realistic experimental conditions.
  5. [§5–6, Data and Code Availability] Code and data are described as available 'upon reasonable request' with a repository 'added upon publication.' Because the paper's comparison cannot be independently reproduced without these, please make the code and at least the experimental and synthetic benchmark data publicly available with the paper.

Circularity Check

1 steps flagged · score 4.0 of 10

Experimental accuracy claim is partly circular: the synthetic-data generator was tuned on the same oxidized p(g3T2) dataset used as the benchmark, and the experimental comparison has no independent ground truth.

  1. fitted input called prediction [Methods §2 (synthetic data generation) / Results §3 (Fig. 4, experimental benchmark)]
    "The generation parameters were initially tuned with reference to experimental data from an oxidized p(g3T2) dataset, an archetypal OMIEC that has been well-studied. ... We used our model to analyze peak detection in a 4DSTEM dataset recorded from an oxidized p(g3T2) sample."

    The simulator producing all training labels and synthetic F1 curves was tuned to the same oxidized p(g3T2) data later used as the experimental benchmark, so that benchmark is not independent. The reported ML advantage measures how well the generator encoded this dataset's peak morphologies, not an external ability to find true peaks. No independent ground-truth labels exist for p(g3T2); the comparison uses coverage counts ('99.65%' vs '85.16%' of probe positions) and 'inspected by eye' checks, which cannot separate true detections from false positives. The extra reduced p(g3T2) dataset is same chemistry, and PB2T-TEG is also only visually assessed. Hence the central experimental accuracy claim reduces in part to the generator's own tuning.

full rationale

The U-Net training and the synthetic F1 validation are self-contained: the model is trained on 500,000 generated images and tested on separately generated images, with explicit F1 metrics (Fig. 3) and an unambiguous dose/intensity dependence. Equation (1) is an independent counting relation, not a fit to the target claim. No uniqueness theorem or load-bearing self-citation chain is used; citations to py4DSTEM and prior OMIEC work are contextual. The only meaningful circularity is in the experimental benchmark: the synthetic-data generator was 'initially tuned with reference to experimental data from an oxidized p(g3T2) dataset,' and the headline experimental comparison is then made on an oxidized p(g3T2) dataset, with no independent labels and only visual verification. This does not make the whole derivation circular, but it makes the central 'outperforms correlative algorithms in almost all cases' claim partially dependent on the same data used to calibrate the training distribution. A score of 4 reflects one significant benchmark-contamination circularity while the core synthetic-data methodology remains independent.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The paper adds a learned model rather than a physical derivation. Its main inputs are hand-tuned synthetic generator parameters, U-Net hyperparameters, and label-construction choices. The most consequential free parameter is the calibration of the generator to the same experimental dataset used as the headline benchmark, which underwrites the experimental comparison.

free parameters (4)
  • Synthetic data generator parameter ranges = tuned to p(g3T2), then expanded; exact values not given
    Peak intensity/position/width distributions, recipe probabilities, and Poisson weights are hand-set and calibrated on the same material used for the main experimental benchmark.
  • Detection threshold = 0.5 (default)
    The single user-set threshold controls the precision/recall tradeoff; default 0.5 is reported sufficient for all cases tested, but it is a free knob.
  • U-Net architecture hyperparameters = 4 layers, 3 convs/layer, 32 starting filters, 3×3 kernel, 200 epochs (best at epoch 191)
    Chosen without a systematic architecture search; model capacity and training length affect the central results.
  • Label construction parameters = 3-pixel Gaussian truncation, 8-pixel intensity circle radius, Voronoi overlap rule
    Hand-chosen target label geometry affects both the training signal and the F1 evaluation metrics.
assumptions (5)
  • domain assumption Electron detection noise is Poisson-distributed and the synthetic generator's recipes span the real diffraction variations.
    Section 2: generator uses 'rates (Poisson statistics)' and recipes including lamellar, backbone, π-π, ice, contamination, dumbbell; if real patterns include unmodeled harmonics or other morphologies, the model will miss them (the paper itself notes harmonics are an edge case).
  • domain assumption A U-Net trained purely on synthetic data generalizes to experimental 4DSTEM data without harmful domain shift.
    Core transfer assumption underlying the experimental results in Section 3; only indirect evidence (synthetic F1 and visual inspection) supports it.
  • domain assumption The three reflection classes (lamellar, backbone, π-π) and their morphologies are sufficient and correctly labeled in simulated data.
    Section 1 and 2 build the method around these three classes; if a polymer produces other peak shapes, model coverage degrades.
  • ad hoc to paper Visual inspection by the authors with surrounding probe-position context is a valid ground truth for judging faint peaks.
    Fig. 4 caption: 'Faint or uncertain peaks... were judged by the authors using surrounding probe positions for context.' This evaluation assumption is specific to this paper and not independently verified.
  • standard math Standard U-Net training and local-maximum/subpixel peak refinement behave as intended.
    Section 2: uses a standard architecture and conventional peak post-processing; no formal verification is provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Mapping Order in Semicrystalline Polymers using Machine Learning of Nanobeam Electron Diffraction." pith.science (2026). https://pith.science/paper/B3QNNIIS

@misc{pith2026260716570,
  author       = {Pith},
  title        = {Pith review of: Mapping Order in Semicrystalline Polymers using Machine Learning of Nanobeam Electron Diffraction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/B3QNNIIS}},
  note         = {Machine review of arXiv:2607.16570}
}
read the original abstract

Organic mixed ionic electronic conductors (OMIECs) are a promising class of polymer materials for applications spanning neuromorphic computation to energy efficient electronics and bioelectronics. Despite being highly tunable, the relationship between structural features and key performance properties such as charge carrier mobility is poorly understood. Scanning nanodiffraction in the transmission electron microscope (TEM) is a powerful probe for elucidating this structure-property relationship, but produces large, noisy datasets that are difficult to interpret because polymer reflections exhibit several distinct morphologies. To address the complexity, we trained a machine learning (ML) model to detect these polymer diffraction peaks and their intensities from synthetic data. Compared to correlative peak detection algorithms, the conventional method for analyzing nanobeam 4D scanning transmission electron microscopy (4DSTEM) data, we show that the ML model is significantly faster and outperforms correlative algorithms in almost all cases, opening up the possibility of near-live visualization of 4DSTEM experiments.

Figures

Figures reproduced from arXiv: 2607.16570 by the authors.

Figure 1
Figure 1. Schematic of a 4DSTEM experiment on a semicrystalline polymer. Each panel shows a representative crystallite orientation and the corresponding synthetic diffraction pattern when the indicated Bragg condition is satisfied: (a) lamellar stacking, (b) backbone periodicity, and (c) π-π stacking. initially tuned with reference to experimental data from an oxidized p(g3T2) dataset, an archetypal OMIEC that has been well-s… view at source ↗
Figure 2
Figure 2. Overview of the ML-based peak detection algorithm. (a) Example synthetic training data used to train the model. The ideal diffraction pattern with relevant geometry is shown and indicated. The augmented diffraction pattern is fed as input to the model. The position and intensity outputs are then compared to the labels. (b) Schematic of the model U-Net architecture. (c) Example experimental input data and (d) associa… view at source ↗
Figure 3
Figure 3. Validation results on synthetic data. Intensity is normalized to either the central beam intensity or detector saturation, whichever is lower. (a)–(c) F1 curves for lamellar, backbone, and π-π peaks, respectively, all with respect to peak counts. (d)–(f) F1 curves for lamellar, backbone, and π-π peaks, respectively, all with respect to peak intensity. Annular spread units, σφ , are given in degrees. kpr is the probe… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Experimental results on oxidized p(g3T2): ML vs. correlative template matching. Experimental 4DSTEM data from an oxidized p(g3T2) sample were analyzed with both the ML model developed in this work and a correlative peak-detection algorithm, with detected peaks split by…
Figure 5
Figure 5. Figure 5: Orientation map comparison from experimental oxidized p(g3T2) 4DSTEM data. (a)–(c) Orientation maps from peaks detected by correlative template matching and (d)–(f) by the ML model developed in this work, for lamellar, backbone, and π-π peaks, respectively. The inset l…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

39 extracted references · 1 canonical work pages

  1. [1]

    Nat Rev Mater3, 17086 (2018)

    Rivnay, J.et al.Organic electrochemical transistors. Nat Rev Mater3, 17086 (2018)

    2018

  2. [2]

    Wong, W. S. & Salleo, A. (eds.)Flexible Electronics: Materials and Applications, vol. 11 ofElectronic Materials: Science & Technology(Springer US, Boston, MA, 2009)

    2009

  3. [3]

    D., Tybrandt, K., Stavrinidou, E

    Paulsen, B. D., Tybrandt, K., Stavrinidou, E. & Rivnay, J. Organic mixed ionic–electronic conductors.Nat. Mater .19, 13–26 (2020)

    2020

  4. [4]

    Mater 16, 414–418 (2017)

    Van De Burgt, Y .et al.A non-volatile organic electrochemical device as a low-voltage artificial synapse for neuromorphic computing.Nat. Mater 16, 414–418 (2017)

    2017

  5. [5]

    & Rovira, C

    Mas-Torrent, M. & Rovira, C. Role of Molecular Order and Solid-State Structure in Organic Field-Effect Transistors.Chem. Rev.111, 4833–4856 (2011)

    2011

  6. [6]

    Beaujuge, P. M. & Fréchet, J. M. J. Molecular Design and Ordering Effects in π-Functional Materials for Transistor and Solar Cell Applications. J. Am. Chem. Soc.133, 20009–20029 (2011)

    2011

  7. [7]

    & Pei, J

    Yao, Z.-F., Wang, J.-Y . & Pei, J. Control of π–π Stacking via Crystal Engineering in Organic Conjugated Small Molecule Crystals.Cryst. Growth & Des.18, 7–15 (2018)

    2018

  8. [8]

    X.et al.Structure Control of a π-Conjugated Oligothiophene-Based Liquid Crystal for Enhanced Mixed Ion/Electron Transport Characteristics.ACS Nano13, 7665–7675 (2019)

    Dong, B. X.et al.Structure Control of a π-Conjugated Oligothiophene-Based Liquid Crystal for Enhanced Mixed Ion/Electron Transport Characteristics.ACS Nano13, 7665–7675 (2019)

    2019

Show all 39 references
  1. [9]

    Himmelberger, S.CHARGE TRANSPORT IN SEMICONDUCTING POLYMERS: CONNECTING MICROSTRUCTURE TO ELECTRONIC PROPERTIES. Ph.D. thesis, Stanford University, Department of Materials Science and Engineering (2015)

    2015

  2. [10]

    G.et al.A Reversible Structural Phase Transition by Electrochemically-Driven Ion Injection into a Conjugated Polymer.J

    Bischak, C. G.et al.A Reversible Structural Phase Transition by Electrochemically-Driven Ion Injection into a Conjugated Polymer.J. Am. Chem. Soc.142, 7434–7442 (2020)

    2020

  3. [11]

    Libera, M. R. & Egerton, R. F. Advances in the Transmission Electron Microscopy of Polymers. Polym. Rev.50, 321–339 (2010)

    2010

  4. [12]

    Michler, G. H. & Lebek, W. Electron Microscopy of Polymers. InPolymer Morphology, chap. 3, 37–53, DOI: 10.1002/9781118892756.ch3 (John Wiley & Sons, Ltd, 2016)

    2016 doi

  5. [13]

    ACS Macro Lett.10, 1306–1314 (2021)

    Cendra, C.et al.Unraveling the Unconventional Order of a High-Mobility Indacenodithiophene–Benzothiadiazole Copolymer. ACS Macro Lett.10, 1306–1314 (2021)

    2021

  6. [14]

    Four-Dimensional Scanning Transmission Electron Microscopy (4D-STEM): From Scanning Nanodiffraction to Ptychography and Beyond.Microsc Microanal25, 563–582 (2019)

    Ophus, C. Four-Dimensional Scanning Transmission Electron Microscopy (4D-STEM): From Scanning Nanodiffraction to Ptychography and Beyond.Microsc Microanal25, 563–582 (2019)

    2019

  7. [15]

    Quantitative Scanning Transmission Electron Microscopy for Materials Science: Imaging, Diffraction, Spectroscopy, and Tomography.Annu

    Ophus, C. Quantitative Scanning Transmission Electron Microscopy for Materials Science: Imaging, Diffraction, Spectroscopy, and Tomography.Annu. Rev. Mater. Res.53, 105–141 (2023)

    2023

  8. [16]

    C.et al.4D-STEM of Beam-Sensitive Materials.Acc

    Bustillo, K. C.et al.4D-STEM of Beam-Sensitive Materials.Acc. Chem. Res.54, 2543–2551 (2021)

    2021

  9. [17]

    C.et al.Development of Diffraction Scanning Techniques for Beam Sensitive Polymers

    Bustillo, K. C.et al.Development of Diffraction Scanning Techniques for Beam Sensitive Polymers. Microsc Microanal22, 492–493 (2016)

    2016

  10. [18]

    Chen, M.et al.Direct imaging of the crystalline domains and their orientation in the PS-b-PEO block copolymer with 4D-STEM.Macromolecules57, 5629–5638 (2024)

    2024

  11. [19]

    Mater .24, 101–108 (2025)

    Tsarfati, Y .et al.The hierarchical structure of organic mixed ionic–electronic conductors and its evolution in water.Nat. Mater .24, 101–108 (2025)

    2025

  12. [20]

    Mater .18, 860–865 (2019)

    Panova, O.et al.Diffraction imaging of nanocrystalline structures in organic semiconductor molecular thin films.Nat. Mater .18, 860–865 (2019)

    2019

  13. [21]

    Uhlmann, D. R. & Kolbeck, A. G. The Microstructure of Polymeric Materials.Sci. Am. 233, 96–107 (1975). 24949965

    1975

  14. [22]

    Rivnay, J., Mannsfeld, S. C. B., Miller, C. E., Salleo, A. & Toney, M. F. Quantitative Determination of Organic Semiconductor Microstructure from the Molecular to Device Scale.Chem. Rev.112, 5488–5519 (2012). 12/23

    2012

  15. [23]

    Taylor, K. A. & Glaeser, R. M. Electron Microscopy of Frozen Hydrated Biological Specimens.J. Ultrastruct. Res.55, 448–456 (1976)

    1976

  16. [24]

    Taylor, K. A. & Glaeser, R. M. ELECTRON DIFFRACTION OF FROZEN, HYDRATED PROTEIN CRYSTALS.Science(1974)

    1974

  17. [25]

    & Dubochet, J

    Knapek, E. & Dubochet, J. Beam damage to organic material is considerably reduced in cryo-electron microscopy.J. Mol. Biol.141, 147–161 (1980)

    1980

  18. [26]

    iScience25(2022)

    Donohue, J.et al.Cryogenic 4D-STEM analysis of an amorphous-crystalline polymer blend: Combined nanocrystalline and amorphous phase mapping. iScience25(2022)

    2022

  19. [27]

    F., Lazar, S

    Egerton, R. F., Lazar, S. & Libera, M. Delocalized radiation damage in polymers.Micron43, 2–7 (2012)

    2012

  20. [28]

    C., Gammer, C., Ciston, J., Minor, A

    Pekin, T. C., Gammer, C., Ciston, J., Minor, A. M. & Ophus, C. Optimizing disk registration algorithms for nanobeam electron diffraction strain mapping. Ultramicroscopy176, 170–176 (2017)

    2017

  21. [29]

    Rauch, E. F.et al.Automated nanocrystal orientation and phase mapping in the transmission electron microscope on the basis of precession electron diffraction.Zeitschrift für Kristallographie 225, 103–109 (2010)

    2010

  22. [30]

    & Zuo, J.-M

    Meng, Y . & Zuo, J.-M. Improvements in Electron Diffraction Pattern Automatic Indexing Algorithms. The Eur. Phys. J. Appl. Phys.80, 10701 (2017)

    2017

  23. [31]

    Ophus, C.et al.Automated Crystal Orientation Mapping in py4DSTEM using Sparse Correlation Matching.Microanal28, 390–403 (2022)

    2022

  24. [32]

    H.et al.py4DSTEM: A Software Package for Four-Dimensional Scanning Transmission Electron Microscopy Data Analysis

    Savitzky, B. H.et al.py4DSTEM: A Software Package for Four-Dimensional Scanning Transmission Electron Microscopy Data Analysis. Microanal27, 712–743 (2021)

    2021

  25. [33]

    Mater.12, 145 (2026)

    Liu, M.et al.A unified preprocessing framework for high-throughput diffraction pattern analysis.npj Comput. Mater.12, 145 (2026)

    2026

  26. [34]

    Mater8, 254 (2022)

    Munshi, J.et al.Disentangling multiple scattering with deep learning: Application to strain mapping from electron diffraction patterns.npj Comput. Mater8, 254 (2022)

    2022

  27. [35]

    H., Johnstone, D

    Martineau, B. H., Johnstone, D. N., van Helvoort, A. T. J., Midgley, P. A. & Eggeman, A. S. Unsupervised machine learning applied to scanning precession electron diffraction data.Adv Struct Chem Imag5, 3 (2019)

    2019

  28. [36]

    & Zuo, J.-M

    Yuan, R., Zhang, J., He, L. & Zuo, J.-M. Training artificial neural networks for precision orientation and strain mapping using 4D electron diffraction datasets.Ultramicroscopy231, 113256 (2021)

    2021

  29. [37]

    Mater8, 114 (2022)

    Shi, C.et al.Uncovering material deformations via machine learning combined with four-dimensional scanning transmission electron microscopy.npj Comput. Mater8, 114 (2022)

    2022

  30. [38]

    Zintler, A.et al.Machine Learning Assisted Pattern Matching: Insight into Oxide Electronic Device Performance by Phase Determination in 4D-STEM Datasets.Microsc Microanal26, 1908–1909 (2020)

    1908

  31. [39]

    & V oyles, P

    Francis, C. & V oyles, P. M. Clustering characteristic diffraction vectors in 4-D STEM data sets from overlapping structures in nanocrystalline and amorphous materials.Ultramicroscopy267, 114040 (2024). 13/23 Supporting Information Figure S1.Various loss and parameter curves f...

    2024

Pith tools

Reviewed August 1, 2026 · model on record in the stance chip above.