Low Photon Count Phase Retrieval Using Deep Learning
Imaging systems' performance at low light intensity is affected by shot noise, which becomes increasingly strong as the power of the light source decreases. In this Letter, we experimentally demonstrate the use of deep neural networks to recover objects illuminated with weak light and demonstra...
Saved in:
| Published in | Physical review letters Vol. 121; no. 24; p. 243902 |
|---|---|
| Main Authors | , , , |
| Format | Journal Article |
| Language | English |
| Published |
United States
14.12.2018
|
| Online Access | Get more information |
Cover
Loading…
| Abstract | Imaging systems' performance at low light intensity is affected by shot noise, which becomes increasingly strong as the power of the light source decreases. In this Letter, we experimentally demonstrate the use of deep neural networks to recover objects illuminated with weak light and demonstrate better performance than with the classical Gerchberg-Saxton phase retrieval algorithm for equivalent signal over noise ratio. The prior contained in the training image set can be leveraged by the deep neural network to detect features with a signal over noise ratio close to one. We apply this principle to a phase retrieval problem and show successful recovery of the object's most salient features with as little as one photon per detector pixel on average in the illumination beam. We also show that the phase reconstruction is significantly improved by training the neural network with an initial estimate of the object, as opposed to training it with the raw intensity measurement. |
|---|---|
| AbstractList | Imaging systems' performance at low light intensity is affected by shot noise, which becomes increasingly strong as the power of the light source decreases. In this Letter, we experimentally demonstrate the use of deep neural networks to recover objects illuminated with weak light and demonstrate better performance than with the classical Gerchberg-Saxton phase retrieval algorithm for equivalent signal over noise ratio. The prior contained in the training image set can be leveraged by the deep neural network to detect features with a signal over noise ratio close to one. We apply this principle to a phase retrieval problem and show successful recovery of the object's most salient features with as little as one photon per detector pixel on average in the illumination beam. We also show that the phase reconstruction is significantly improved by training the neural network with an initial estimate of the object, as opposed to training it with the raw intensity measurement. |
| Author | Arthur, Kwabena Barbastathis, George Li, Shuai Goy, Alexandre |
| Author_xml | – sequence: 1 givenname: Alexandre surname: Goy fullname: Goy, Alexandre organization: Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA – sequence: 2 givenname: Kwabena surname: Arthur fullname: Arthur, Kwabena organization: Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA – sequence: 3 givenname: Shuai surname: Li fullname: Li, Shuai organization: Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA – sequence: 4 givenname: George surname: Barbastathis fullname: Barbastathis, George organization: Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA |
| BackLink | https://www.ncbi.nlm.nih.gov/pubmed/30608745$$D View this record in MEDLINE/PubMed |
| BookMark | eNo1jl1LwzAUhoMo7kP_wsgfaD0nSdP0UqpOoeAY7nqkzamrbOlossn-vQX15nl5bl6eGbv2vSfGFggpIsiH1e4S1nSuKMYUBaZCyQLEFZsi5EWSI6oJm4XwBQAotLllEwkaTK6yKVNV_81Xuz72npf9ycdRbCC-pjh0dLZ7vgmd_-RPREdekR38aHfsprX7QPd_O2ebl-eP8jWp3pdv5WOVNCqTcaRT4KhGk7WmMWTAtgjKEDagrM5QOeewMQUUWguZE8miNjLXTmlwphFztvj9PZ7qA7ntcegOdrhs__PFD8deSLU |
| CitedBy_id | crossref_primary_10_1038_s41377_020_0302_3 crossref_primary_10_1038_s41377_022_00949_8 crossref_primary_10_1364_OE_498073 crossref_primary_10_1088_1361_6501_ab44d8 crossref_primary_10_1073_pnas_1821378116 crossref_primary_10_1016_j_optlaseng_2022_107196 crossref_primary_10_1109_TCI_2022_3209936 crossref_primary_10_1364_OE_424165 crossref_primary_10_1038_s42256_023_00704_7 crossref_primary_10_1109_TSP_2022_3156014 crossref_primary_10_1364_AO_390256 crossref_primary_10_1109_ACCESS_2020_2978435 crossref_primary_10_1016_j_optlastec_2024_111726 crossref_primary_10_1364_OE_434014 crossref_primary_10_1109_TSP_2022_3203228 crossref_primary_10_1364_PRJ_440123 crossref_primary_10_3390_ai3020020 crossref_primary_10_1080_09500340_2021_1896815 crossref_primary_10_1364_OPTICA_392465 crossref_primary_10_1007_s00340_022_07764_4 crossref_primary_10_1021_acsphotonics_1c00337 crossref_primary_10_1007_s40042_022_00412_2 crossref_primary_10_1098_rsta_2023_0346 crossref_primary_10_1038_s41598_020_76764_1 crossref_primary_10_1364_OE_445498 crossref_primary_10_1364_OPTICA_470712 crossref_primary_10_1063_5_0009493 crossref_primary_10_1364_PRJ_412965 crossref_primary_10_1088_2040_8986_abe18a crossref_primary_10_1002_adpr_202200264 crossref_primary_10_1088_0256_307X_42_3_030205 crossref_primary_10_1063_1_5125252 crossref_primary_10_1063_5_0076022 crossref_primary_10_1364_OE_397790 crossref_primary_10_1038_s42256_022_00584_3 crossref_primary_10_1364_OSAC_383527 crossref_primary_10_1109_TCI_2024_3458418 crossref_primary_10_1038_s41467_023_41496_z crossref_primary_10_1109_JSTQE_2021_3060762 crossref_primary_10_1364_OL_441441 crossref_primary_10_1038_s41566_020_00716_4 crossref_primary_10_1038_s44172_023_00103_1 crossref_primary_10_1364_OL_420388 crossref_primary_10_1063_5_0176874 crossref_primary_10_1016_j_scitotenv_2024_177084 crossref_primary_10_1080_09500340_2021_1977860 crossref_primary_10_1364_OL_440900 crossref_primary_10_3390_s19112462 crossref_primary_10_1364_OPTICA_453748 crossref_primary_10_1117_1_APN_3_5_056002 crossref_primary_10_3390_cells11152394 crossref_primary_10_1364_OL_487000 crossref_primary_10_1038_s41592_023_02041_4 crossref_primary_10_1088_1361_6463_ac2ad4 crossref_primary_10_1364_OE_395000 crossref_primary_10_1016_j_optlaseng_2021_106641 crossref_primary_10_1109_TIP_2024_3445739 crossref_primary_10_1016_j_optlaseng_2020_106449 crossref_primary_10_1109_TPAMI_2021_3138897 crossref_primary_10_1186_s43593_022_00037_9 crossref_primary_10_1364_BOE_395302 crossref_primary_10_1364_JOSAB_489100 crossref_primary_10_1364_OE_27_025560 crossref_primary_10_1364_OE_513556 crossref_primary_10_1364_PRJ_447862 crossref_primary_10_1016_j_imavis_2024_105282 crossref_primary_10_1364_BOE_464367 crossref_primary_10_3938_jkps_76_401 crossref_primary_10_1109_TTHZ_2024_3349482 crossref_primary_10_1364_BOE_444488 crossref_primary_10_1364_OPTICA_374026 crossref_primary_10_1016_j_optlaseng_2021_106619 crossref_primary_10_1007_s12551_019_00554_x crossref_primary_10_1088_2515_7647_abf02c crossref_primary_10_1038_s41377_020_0267_2 crossref_primary_10_1364_OE_381301 crossref_primary_10_1145_3689037 crossref_primary_10_1364_OPTICA_6_000921 crossref_primary_10_1038_s41598_020_62484_z crossref_primary_10_1016_j_optcom_2025_131632 crossref_primary_10_1109_LSP_2019_2935814 crossref_primary_10_1364_OE_518249 crossref_primary_10_1088_1361_6560_acc2aa crossref_primary_10_1038_s41377_022_00714_x crossref_primary_10_1364_OE_390413 crossref_primary_10_1038_s41377_023_01340_x crossref_primary_10_1364_OE_412009 crossref_primary_10_3390_s22218366 crossref_primary_10_1103_PhysRevApplied_19_034005 crossref_primary_10_1063_5_0049111 crossref_primary_10_1103_PhysRevE_102_011001 crossref_primary_10_1038_s41598_021_90312_5 crossref_primary_10_3390_app122211436 crossref_primary_10_37188_lam_2022_013 crossref_primary_10_1088_2515_7620_aba6d1 crossref_primary_10_1021_acs_analchem_3c00274 crossref_primary_10_1103_PhysRevA_99_053419 crossref_primary_10_1364_OE_27_033120 crossref_primary_10_3389_fphy_2021_710351 crossref_primary_10_1016_j_optlaseng_2020_106233 crossref_primary_10_1364_OE_397430 crossref_primary_10_1038_s41377_019_0196_0 crossref_primary_10_1038_s41377_023_01181_8 crossref_primary_10_1364_OL_411564 crossref_primary_10_1021_acsphotonics_4c01863 crossref_primary_10_1364_OE_520066 crossref_primary_10_1364_OL_397717 crossref_primary_10_1088_2399_6528_ac016a crossref_primary_10_1364_OE_461367 crossref_primary_10_1364_OE_520788 crossref_primary_10_1103_PhysRevResearch_3_043066 crossref_primary_10_1364_AO_544424 crossref_primary_10_1364_PRJ_415590 crossref_primary_10_1038_s41586_020_2973_6 crossref_primary_10_1016_j_optlaseng_2022_107010 crossref_primary_10_1364_OE_395204 crossref_primary_10_29026_oea_2024_230034 crossref_primary_10_1364_OPTICA_506572 crossref_primary_10_3390_rs12081289 crossref_primary_10_1117_1_OE_63_7_073102 crossref_primary_10_1063_5_0014725 crossref_primary_10_1021_acsphotonics_1c01365 crossref_primary_10_1364_OE_460208 crossref_primary_10_1364_PRJ_416551 crossref_primary_10_1016_j_pacs_2024_100645 crossref_primary_10_1103_PhysRevLett_123_183902 crossref_primary_10_1103_PhysRevA_101_063417 crossref_primary_10_1038_s41524_025_01569_7 crossref_primary_10_1364_OL_411228 crossref_primary_10_1109_JPROC_2019_2949575 crossref_primary_10_1364_JOSAA_440610 crossref_primary_10_1002_lpor_202300939 crossref_primary_10_1038_s41377_021_00512_x crossref_primary_10_1364_OE_528687 crossref_primary_10_1515_nanoph_2021_0601 |
| ContentType | Journal Article |
| DBID | NPM |
| DOI | 10.1103/PhysRevLett.121.243902 |
| DatabaseName | PubMed |
| DatabaseTitle | PubMed |
| DatabaseTitleList | PubMed |
| Database_xml | – sequence: 1 dbid: NPM name: PubMed url: https://proxy.k.utb.cz/login?url=http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed sourceTypes: Index Database |
| DeliveryMethod | no_fulltext_linktorsrc |
| Discipline | Physics |
| EISSN | 1079-7114 |
| ExternalDocumentID | 30608745 |
| Genre | Journal Article |
| GroupedDBID | --- -DZ -~X 123 2-P 29O 3MX 5VS 85S ACBEA ACGFO ACNCT AENEX AEQTI AFFNX AFGMR AGDNE AJQPL ALMA_UNASSIGNED_HOLDINGS APKKM AUAIK CS3 D0L DU5 EBS EJD ER. F5P MVM N9A NPBMV NPM OK1 P2P ROL S7W SJN TN5 UBE WH7 XSW YNT ZPR ~02 |
| ID | FETCH-LOGICAL-c453t-c4d40deb185f8c8e80af1048e1c04a6514ddd1c890966237ee39b8376d460d8c2 |
| IngestDate | Thu Apr 03 06:59:48 EDT 2025 |
| IsDoiOpenAccess | false |
| IsOpenAccess | true |
| IsPeerReviewed | true |
| IsScholarly | true |
| Issue | 24 |
| Language | English |
| LinkModel | OpenURL |
| MergedId | FETCHMERGED-LOGICAL-c453t-c4d40deb185f8c8e80af1048e1c04a6514ddd1c890966237ee39b8376d460d8c2 |
| OpenAccessLink | http://link.aps.org/pdf/10.1103/PhysRevLett.121.243902 |
| PMID | 30608745 |
| ParticipantIDs | pubmed_primary_30608745 |
| PublicationCentury | 2000 |
| PublicationDate | 2018-Dec-14 |
| PublicationDateYYYYMMDD | 2018-12-14 |
| PublicationDate_xml | – month: 12 year: 2018 text: 2018-Dec-14 day: 14 |
| PublicationDecade | 2010 |
| PublicationPlace | United States |
| PublicationPlace_xml | – name: United States |
| PublicationTitle | Physical review letters |
| PublicationTitleAlternate | Phys Rev Lett |
| PublicationYear | 2018 |
| SSID | ssj0001268 |
| Score | 2.6419384 |
| Snippet | Imaging systems' performance at low light intensity is affected by shot noise, which becomes increasingly strong as the power of the light source decreases. In... |
| SourceID | pubmed |
| SourceType | Index Database |
| StartPage | 243902 |
| Title | Low Photon Count Phase Retrieval Using Deep Learning |
| URI | https://www.ncbi.nlm.nih.gov/pubmed/30608745 |
| Volume | 121 |
| hasFullText | |
| inHoldings | 1 |
| isFullTextHit | |
| isPrint | |
| link | http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwnV3dS8MwEA9-oOxF_P6WPPgmnemadumj-DVERXQD36RNLk7QbcxOwb_eS9J2Y078eAlrQrM29-vl7rjfhZD9iCfaZtTIlAmPh6C9OADpaa5CNHdBp9Jwh6-uo0aLX9yH98NcVcsuydKq_JjIK_mPVLEP5WpYsn-QbDkpduBvlC-2KGFsfyXjS8Nja3dNZQzDLc_wAjclXDJzShb-0YFLCDgB6BWFVB9HrdGbQkg5geXZcntKK_vcxelyDkx_BBxZe-ByMd6TFDqlZr-0qQF37UHyNIyQ9nGjzMpzj10MfjTW4AuTt-E4nlVw-pHVY6_u532FAnUc5xwpNT6qD9HesYzqCaqamZIR5kVv4c1wl0yhi-rXG3DJey9WgOjdMFOd_-fRsRLaxdA0mUZnwpyOakI6lSIQF4mcOo6PdDj5gSpkvphkzP-wdkhzkSzkDgQ9cmhYIlPQWSZzTpKvK4QjJqjDBLWYoBYTtMQEtZigBhO0wMQqaZ2dNo8bXn4yhid5GGTYKs4UbrMi1EIKEAy_ONTF4EvGkwiNYKWUL0WMDirat3WAIE4F7iWKR0wJWVsjM51uBzYITUApHQpA1xLnjpJYgqrVwQ8CGWim1SZZd6_70HPlTx6Khdj6dmSbVIbo2SGzGr832EXjLUv37OJ_AmxwQ90 |
| linkProvider | National Library of Medicine |
| openUrl | ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Low+Photon+Count+Phase+Retrieval+Using+Deep+Learning&rft.jtitle=Physical+review+letters&rft.au=Goy%2C+Alexandre&rft.au=Arthur%2C+Kwabena&rft.au=Li%2C+Shuai&rft.au=Barbastathis%2C+George&rft.date=2018-12-14&rft.eissn=1079-7114&rft.volume=121&rft.issue=24&rft.spage=243902&rft_id=info:doi/10.1103%2FPhysRevLett.121.243902&rft_id=info%3Apmid%2F30608745&rft_id=info%3Apmid%2F30608745&rft.externalDocID=30608745 |