TEsoNet: knowledge transfer in surgical phase recognition from laparoscopic sleeve gastrectomy to the laparoscopic part of Ivor-Lewis esophagectomy.

BACKGROUND: Surgical phase recognition using computer vision presents an essential requirement for artificial intelligence-assisted analysis of surgical workflow. Its performance is heavily dependent on large amounts of annotated video data, which remain a limited resource, especially concerning highly specialized procedures. Knowledge transfer from common to more complex procedures can promote data efficiency. Phase recognition models trained on large, readily available datasets may be extrapolated and transferred to smaller datasets of different procedures to improve generalizability. The conditions under which transfer learning is appropriate and feasible remain to be established. METHODS: We defined ten operative phases for the laparoscopic part of Ivor-Lewis Esophagectomy through expert consensus. A dataset of 40 videos was annotated accordingly. The knowledge transfer capability of an established model architecture for phase recognition (CNN + LSTM) was adapted to generate a "Transferal Esophagectomy Network" (TEsoNet) for co-training and transfer learning from laparoscopic Sleeve Gastrectomy to the laparoscopic part of Ivor-Lewis Esophagectomy, exploring different training set compositions and training weights. RESULTS: The explored model architecture is capable of accurate phase detection in complex procedures, such as Esophagectomy, even with low quantities of training data. Knowledge transfer between two upper gastrointestinal procedures is feasible and achieves reasonable accuracy with respect to operative phases with high procedural overlap. CONCLUSION: Robust phase recognition models can achieve reasonable yet phase-specific accuracy through transfer learning and co-training between two related procedures, even when exposed to small amounts of training data of the target procedure. Further exploration is required to determine appropriate data amounts, key characteristics of the training procedure and temporal annotation methods required for successful transferal phase recognition. Transfer learning across different procedures addressing small datasets may increase data efficiency. Finally, to enable the surgical application of AI for intraoperative risk mitigation, coverage of rare, specialized procedures needs to be explored.

How effective is laparoscopic redo-antireflux surgery?

BACKGROUND: The failure-rate after primary antireflux surgery ranges from 3 to 30%. Reasons for failures are multifactorial. The aim of this study is to gain insight into the complex reasons for, and management of, failure after antireflux surgery. METHODS: Patients were selected for redo-surgery after a diagnostic workup consisting of history and physical examination, upper gastrointestinal endoscopy, quality-of-life assessment, screening for somatoform disorders, esophageal manometry, 24-hour-pH-impedance monitoring, and selective radiographic studies such as Barium-sandwich for esophageal passage and delayed gastric emptying. Perioperative and follow-up data were compiled between 2004 and 2017. RESULTS: In total, 578 datasets were analyzed. The patient cohort undergoing a first redo-procedure (n = 401) consisted of 36 patients after in-house primary LF and 365 external referrals (mean age: 62.1 years [25-87]; mean BMI 26 [20-34]). The majority of patients underwent a repeated total or partial laparoscopic fundoplication. Major reasons for failure were migration and insufficient mobilization during the primary operation. With each increasing number of required redo-operations, the complexity of the redo-procedure itself increased, follow-up quality-of-life decreased (GIQLI: 106; 101; and 100), and complication rate increased (intraoperative: 6,4-10%; postoperative: 4,5-19%/first to third redo). After three redo-operations, resections were frequently necessary (morbidity: 42%). CONCLUSIONS: Providing a careful patient selection, primary redo-antireflux procedures have proven to be highly successful. It is often the final chance for a satisfying result may be achieved upon performing a second redo-procedure. A third revision may solve critical problems, such as severe pain and/or inadequate nutritional intake. When resection is required, quality of life cannot be entirely normalized.

Impact of the D genome and quantitative trait loci on quantitative traits in a spring durum by spring bread wheat cross.

The impact of the D genome and QTL in the A and B genomes on agronomic performance of hexaploid wheat and tetraploid durum was determined using novel recombinant inbred line populations derived from interploid crosses. Genetic differences between common hexaploid (6X) bread wheat (Triticum aestivum, 2n = 6x = 42, genome, AABBDD) and tetraploid (4X) durum wheat (T. turgidum subsp. durum, 2n = 4x = 28, genome, AABB) may exist due to effects of the D genome and allelic differences at loci in the A and B genomes. Previous work allowed identification of a 6X by 4X cross combination that resulted in a large number of fertile recombinant progeny at both ploidy levels. In this study, interspecific recombinant inbred line populations at both 4X and 6X ploidy with 88 and 117 individuals, respectively, were developed from a cross between Choteau spring wheat (6X) and Mountrail durum wheat (4X). The presence of the D genome in the 6X population resulted in increased yield, tiller number, kernel weight, and kernel size, as well as a decrease in stem solidness, test weight and seed per spike. Similar results were found with a second RIL population containing 152 lines from 18 additional 6X by 4X crosses. Several QTL for agronomic and quality traits were identified in both the 4X and 6X populations. Although negatively impacted by the lack of the D genome, kernel weight in Mountrail (4X) was higher than Choteau (6X) due to positive alleles from Mountrail on chromosomes 3B and 7A. These and other favorable alleles may be useful for introgression between ploidy levels.

Direct Polymerase Chain Reaction-Based Detection of Cercospora beticola in Field Soils.

Cercospora beticola, the causal agent of Cercospora leaf spot of sugar beet, survives as pseudostromata in infected sugar beet residues in the soil. Under optimal conditions, overwintering propagules germinate and produce conidia that are dispersed as primary inoculum to initiate infection in sugar beet. We developed a polymerase chain reaction (PCR) technique for rapid detection of C. beticola in field soils. Total DNA was first isolated from soil amended with C. beticola culture using the PowerSoil DNA Kit. The purified DNA was subjected to PCR in Extract-N-Amp PCR mix with CBACTIN primers over 35 cycles. The amplified products were resolved and compared by electrophoresis in 1% agarose gels. The PCR fragment size of C. beticola from the amended field soil correlated in size with the amplicon from control C. beticola culture DNA extract. Additionally, sample soils were collected from sugar beet fields near Sidney, MT and Foxholm, ND. Total DNA was extracted from the samples and subjected to PCR and resolved as previously described. The amplicons were purified from the gels and subjected to BigDye Terminator Cycle sequencing. All sequences from field soils samples, C. beticola-amended field soil, and pure culture were compared by alignment with a C. beticola actin gene sequence from GenBank. The result of the alignment confirmed the amplicons as products from C. beticola. Rapid screening for the presence of C. beticola in the soil by PCR will improve research capabilities in biological control, disease forecasting, and management of this very important sugar beet pathogen.

Watershed nitrogen and mercury geochemical fluxes integrate landscape factors in long-term research watersheds at Acadia National Park, Maine, USA.

This paper is an overview of this special issue devoted to watershed research in Acadia National Park (Acadia NP). The papers address components of an integrated research program on two upland watersheds at Acadia NP, USA (44 degrees 20' N latitude; 68 degrees 15' E longitude). These watersheds were instrumented in 1998 to provide a long-term foundation for regional ecological and watershed research. The research was initiated as part of EPA/NPS PRIMENet (Park Research and Intensive Monitoring of Ecosystems Network), a system of UV-monitoring stations and long-term watershed research sites located in US national parks. The initial goals at Acadia NP were to address research questions about mercury, acid rain, and nitrogen saturation developed from prior research. The project design was based on natural differences in forests and soils induced by an intense wildfire in one watershed in 1947. There is no evidence of fire in the reference watershed for several hundred years. We are testing hypotheses about controls on surface water chemistry, and bioavailability of contaminants in the contrasting watersheds. The unburned 47-ha Hadlock Brook watershed is 70% spruce-fir mature conifer forest. In contrast, burned 32-ha Cadillac Brook watershed, 4 km northeast of the Hadlock watershed, is 20% regenerating mixed northern hardwoods and 60% shrub/rocky balds. Differences in atmospheric deposition are controlled primarily by forest stand composition and age. The watersheds are gauged and have water chemistry stations at 122 m (Cadillac) and 137 m (Hadlock); watershed maximum elevations are 468 and 380 m, respectively. The stream water chemistry patterns reflect, in part, the legacy of the intense fire, which, in turn, controls differences in forest vegetation and soil characteristics. These factors result in higher nitrogen and mercury flux from the unburned watershed, reflecting differences in atmospheric deposition, contrasting ecosystem pools of nitrogen and mercury, and inferred differences in internal cycling and bioavailabilty.